Warwick Graduate
Conference
in the
Philosophy of Mind
27–28 November 2004
organised by
Jenny Booth, Alli Mitchell and Gloria Ayob
www.warwick.ac.uk/go/mindconf
in association with the
Consciousness and Self-Consciousness
Research Centre
www.warwick.ac.uk/go/consciousnesscentre
Department of Philosophy
1
Contents
Social Externalism, Segal and Contradictory Intuitions
Alex Buckley................................................................................... 3
Transcendental Arguments in the Philosophy of Mind
Mathilde Byskov Jakobsen............................................................17
Conceptualism and The (Supposed) Non-Transitivity of
Colour Indiscriminability
Charles Pelling ..............................................................................29
Relational Looking and the Reversibility Constraint
Richard Price.................................................................................39
Perception Involving Action
Susanna Schellenberg ..................................................................49
The Rationality of Deductive Inference
Declan Smithies ............................................................................63
Social Externalism, Segal and Contradictory Intuitions
Alex Buckley
Department of Philosophy, University of Leeds
phl1ajb@leeds.ac.uk
In his recent publication, A Slim Book About Narrow Content (2000)1, Gabriel
Segal has attempted to argue that the social externalism of Tyler Burge is
inadequate as an account of the fundamental nature of folk psychological
explanation. The first part of this paper will comprise a rebuttal of this
argument, on the basis that Segal assumes the truth of a disputed premise. In
part two I will discuss the ramifications of the rebuttal of the argument. Briefly:
the rebuttal leaves both internalists and externalists with a messy situation
concerning our intuitions, and the natural way to create some order can raise
problems for the social externalist. So the position expressed in this paper as
a whole is that whilst Segal’s argument, as it stands, fails to seriously call
social externalism into question, it does indirectly mount a serious challenge.
Finally, I shall outline a response the externalist can make.
I – Segal and Social Externalism
1.
Segal’s argument
Segal’s argument is intended to show that a certain form of social externalism,
attributed to Tyler Burge, cannot give an acceptable account of belief
ascription when faced with certain crucial puzzle cases. “Consumerism” (the
label is from Kaplan 1989) is the theory that one’s words have the meaning
they do because of the linguistic conventions of the community of which one is
a member. One’s ability to be a functioning (competent and appropriately
deferential) member of that linguistic community endows one’s words with
certain properties, and affects the beliefs and concepts one may correctly be
attributed.
Words come to us prepackaged with a semantic value. If we are to use
those words, the words we have received, the words of our linguistic
community, then we must defer to their meaning…To use language as
language, to express something, requires an intentional act. But the
intention that is required involves the typical consumer’s attitude of
compliance, not the producer’s assertiveness. (Kaplan 1989, 602)
This position is exemplified in the intended conclusion of Burge’s well-known
Twin-Earth thought experiments: intrinsic duplicates can be in different mental
states, purely because of a difference in the linguistic practices of their
communities. What beliefs a person has is not fixed by what is going on in
their heads – social environment plays a determining role. Whilst
1
All unattributed references are to this work.
3
acknowledging the intuitions that support such a conclusion, Segal believes it
is a mistake to rely on them:
There is, indeed, a tendency to think that…Alf believes that he has
arthritis in his thigh, while Twin Alf does not. The mistake is to take this
tendency too seriously and then to draw from it a general conclusion
about the extension of concepts. (122)
The scenario Segal claims to demonstrate the inadequacy of consumerism is
a variation on Burge’s own thought-experiment. “Arthritis” and “inflammation of
the joints” are synonymous, and so anyone, and in particular any expert, who
knows enough about medical ailments to know that they are synonymous will
express the same concept with the two expressions. Alf is not such an expert,
and is not aware of the synonymy, and actually thinks he has arthritis in his
thigh. Not only, therefore, is he not aware of the synonymy of “arthritis” and
“inflammation of the joints”, he would deny that, if asked. “Alf therefore has
two different concepts that he expresses by ‘arthritis’ and ‘inflammation of the
joints’” (65). Where the experts have one concept, Alf has two. But given the
assumptions that a) Alf has some minimal competence with the term
“arthritis”, and can use “inflammation of the joints” to mean what it should
mean, and b) Alf would certainly defer to a room full of arthritologists,
consumerism tells us that Alf gets to express the same concept/s with the
expressions as the experts do. But plainly, Segal argues, he doesn’t. The
“prepackaged semantic value” of “arthritis” and “inflammation of the joints” is
the same, and yet Alf clearly attaches different semantic values to the
expressions. So what consumerism tells us is false. Despite his competence
Alf has his own idiosyncratic concept that he expresses with the term
“arthritis”, that he does not share with the experts.
It is clear that the crucial step in Segal’s argument is the claim that by
not realising that arthritis is just an inflammation of the joint, Alf comes to
express different concepts with the expressions. The support for this step is
not explicit in Segal’s exposition of the argument, but we can plausibly say
that it consists in the following thought: Alf will assent to the sentence “I have
arthritis in my thigh”, and will assent to the sentence “I do not have an
inflammation of the joint in my thigh”. Alf is, by presumption, rational, and
would not assent to both “a is F” and “a is not F”, and so Alf uses the
expressions “arthritis” and “inflammation of the joint” to express different
concepts. Whilst this key claim is implicit and unsupported in Segal (2000), it
is cited in a different context:
Fregean Principle of Difference: If a subject, s, rationally assents to P(t1)
and dissents from or abstains on the truth value of P(t2) then t1 and t2
have different meanings in s’s idiolect and s associates different
concepts with them. (2003, 417)
With such a principle included, Segal’s argument consists in the claim that the
following premises lead to contradiction.
Consumerism: if a subject is minimally competent and deferential with an
expression then they express the same concept with that expression as
experts.
4
Synonymy: experts in the use of synonymous expressions express the
same concept with these expressions.
Fregean Principle of Difference (FPD): if a rational subject sincerely and
on reflection assents to “a is F” and “b is not F” then the subject
expresses different concepts with the expressions “a” and “b”.
In response to Burge’s (1978) claim that synonyms are not inter-substitutable
in belief attributions, and thus denial of Synonymy above, Segal formulates a
version of the argument without that premise: a deferential and minimally
competent subject thinks that one term, for instance “coffee”, expresses two
different concepts, and assents to “all coffee [type 1] comes from exotic
countries” and “all coffee [type 2] does not come from exotic countries”. By
FPD, the subject expresses two concepts with the term “coffee”, where the
experts surely express one. But again, given the minimal competence and
deference that the subject displays despite misunderstanding the term,
Consumerism gives us a conflicting answer.
Segal’s argument is thus that FPD and Consumerism together entail a
contradiction. The fact that FPD is implicit and unsupported is therefore
problematic. In another place he claims that it is a “[n]ormal, roughly Fregean,
criteri[on] for the individuation of meaning” (2003, 416, my emphasis). I
suggest that to view Segal’s argument as plausible, we should charitably
interpret him as claiming that the principle is an intuitively supported rule of
normal practice, and that Consumerism as a philosophical theory should
therefore be abandoned.
Consumerism as characterised above is a much simplified version of
Burge’s full account of belief and concept attribution2, but on certain
understandings of the key terms “competence” and “concept”, Burge would
endorse it fully. The reading of “competence” under which Burge would
endorse Consumerism is the same minimalist reading that is used by Segal.
The readings of “concept” that a) Segal is working with and b) under which
Burge would endorse Consumerism, are more complex, and will be the focus
of later discussion.
We should note that on Burge’s account the reference to “experts” is
inessential. Burge holds that a competent speaker defers to their linguistic
community, whether or not there are experts for all particular terms used by
that community. So we can say that through their competence the speaker
expresses the standardly expressed concepts. Given that experts are
competent speakers, Burgean Consumerism is a broader thesis than the
consumerist principle stated above – the former entails the latter. Construed in
that way, Consumerism is prima facie similar to the Disquotational Principle of
Kripke (1979):
DP: if a normal English speaker sincerely assents to “p” then they
believe that p.
2
As given primarily in (1979), (1986) and (1993).
5
2.
Objection to Segal’s argument
Under the charitable interpretation suggested above, Segal’s argument relies
for its force on the strength of the claim that the Fregean Principle of
Difference is intuitively powerful (maybe even obvious) whilst Consumerism is
a controversial philosophical theory.
However, in employing our intuitive reactions to the famous Twin-Earth
thought-experiments as support for their thesis, what externalists claim is that
those intuitions reveal that our normal procedures for ascribing beliefs are
deeply socially externalist. Not only that, but many externalists also claim that
the truth of social externalism, as a principle of normal practice, casts
significant doubt on principles such as FPD, which rely heavily on the
controversial claim that we have privileged access to the contents of our
beliefs. That is, some externalists claim that principles such as FPD depend
on a philosophical theory, most famously formulated by Descartes, about the
extent to which our mental life is transparent to us.
Given all this, it is possible that a social externalist may propose the very
same argument as Segal, showing that FPD and Consumerism contradict,
and draw the exact opposite conclusion: FPD, and its philosophical supports,
must be abandoned. One example of such an argument is that put forward by
Joseph Owens. He begins by accepting, like Segal, that FPD conflicts with
social externalist principles such as Consumerism and the Disquotational
Principle. He then highlights the extent to which our normal practice of belief
ascription conforms to social externalism: if one examines normal practice, the
Disquotational Principle is revealed as fundamental to the way in which we
ascribe beliefs.
It is hard to see how someone could reasonably dissent from [the
Disquotational Principle], suitably qualified; it seems to be a part of our
customary understanding of what is involved in sincerely assenting to a
sentence one understands. Kripke goes so far as to say that [it] appears
to be a self-evident truth. (1989, 291)
If one looks at The Fregean Principle of Difference, the situation is different –
what is certainly undeniable is that no rational person would sincerely and
knowingly assent to contradictory beliefs. But FPD requires more than that, it
requires that a rational subject will not sincerely and on reflection assent to
contradictory beliefs. To move to this stronger principle requires the claim that
through reflection, a subject has the “more primitive ability to introspectively
determine sameness and difference in belief” (1992, 158). Owens argues that
this principle is far from intuitively obvious, and situates it firmly within a school
of philosophical thought, calling it the “Cartesian model of access” (1992,
147).
Having claimed that the Disquotational Principle is supported by our
intuitive understanding of normal practice, and that FPD ultimately relies on a
(controversial) Cartesian philosophical thesis, Owens rejects FPD on those
grounds. The intuitive force of the Disquotational Principle is such that if the
Cartesian model of access contradicts it, then so much the worse for that:
These principles [such as the Disquotational Principle] serve to “define”
or at least delimit a non-Cartesian notion of belief, and to the extent that
we operate with them we ought to abandon the conception of ourselves
6
as having some simple introspective way of discerning contradictions
among our beliefs. (Owens 1989, 295)
In summary, Owens accepts two of the claims that Segal appears to make: a)
that FPD and social externalism contradict, and b) that if a controversial
philosophical claim contradicts our intuitions about normal practice, then it
should be rejected. However, whereas Segal claims that social externalism is
controversial philosophy and FPD is obviously true, Owens ultimately makes
exactly the opposite claim. Their arguments share a common structure and
some common premises, but give mirrored conclusions. For Segal, social
externalism is a philosophical thesis whilst FPD is a principle of normal
practice. For Owens, FPD is a philosophical thesis whilst social externalism is
a part of normal practice.
Segal’s argument, as presented by me, relies almost entirely on the
intuitive force of the Fregean Principle of Difference: on the claim that it is so
central to the way we think about belief, that a principle such as Consumerism
that conflicts with it must be discarded. But what is clear from arguments such
as Joseph Owens’ is that many externalists deny the intuitive force of FPD. In
fact, Owens makes a persuasive case that an acceptance of the force of the
Twin-Earth thought-experiments can lead directly to a denial of the force of
FPD. Segal’s argument ultimately begs the question against the externalist, by
assuming as intuitively obvious a principle regarded as controversial by many
people. Consumerist is not seriously threatened at this point.
II – Ramifications Of Our Rebuttal of Segal
3.
Messy intuitions
While this completes my discussion of the argument presented by Segal, it is
clearly not any sort of satisfactory conclusion. Segal relies in his argument on
the Fregean Principle of Difference. Given that he does not explicitly argue for
it, I have suggested that the most charitable interpretation is that Segal
believes that principle to be an intuitively supported principle of normal
practice, that we shall accept if we reflect on our practices of belief ascription.
However, some externalists, and I have used Owens as an example, explicitly
claim that FPD is revealed as resting on controversial philosophical
assumptions, by the strong intuitions that support the claim of social
externalism to be a principle of normal practice. The situation thus seems to
be that we have two principles, FPD and Consumerism, that are claimed to be
supported by intuition as principles of normal practice. And those two
principles contradict in certain key puzzle cases. Such a situation is
undoubtedly messy, and is downright unacceptable on some plausible
assumptions.
Firstly, we want our intuitions about normal practice to be good guides to
normal practice. That is not to claim that our prima facie reflections on our
practices of belief ascription are always correct, but that extended
philosophical analysis can deliver accurate information about the ways in
which we attribute mental states. Secondly, we want folk psychology and
normal practice to broadly be a good guide to the truth about beliefs and
7
mental states. This is clearly controversial, but I am content to say that
anyone who rejects this assumption would probably not find the issues being
discussed here of paramount importance in the first place. Thirdly, and finally,
we take there to be a fact of the matter about the beliefs and other mental
states which a subject is in. So if we assume that the actual state of affairs is
not contradictory and that folk psychology is a good guide to that state of
affairs, then we should be highly reluctant to suffer a contradiction in our
intuitions about normal practice.
4.
A natural way out
Consumerism and the Fregean Principle of Difference both tell us what to say
about the concepts that a subject expresses with their words when they make
utterances. We can say therefore that the principles, and the intuitions that
support them, contradict only insofar as they are concerned with the same
kind of concept. This observation inspires an obvious solution to the problem
of their apparent conflict: place the blame on the notorious flexibility of the
term “concept”, and say that both Consumerism and FPD are parts of normal
practice, but that they concern different types of concept. This is the familiar
“dual-content” approach. 3
We have at hand a natural and intuitive dual-content approach, in the
comments made by Tyler Burge himself. I employ his version as it is
reasonably straightforward and allows us to highlight clearly the crucial points
of debate, and also the contentious issues will turn on what is common to
dual-content views, not on the peculiarities of the model.
Burge’s distinction is between, in normal terms, a) the (roughly, public or
linguistic) concept that a subject expresses with a term, and b) the (roughly,
private and subjective) explication of the concept that on reflection the subject
would give: the distinction is between the concept, and the conception of the
concept. The concept is the translational meaning of the term, the
translational concept, and the conception of the concept is the explicational
meaning of the term, the explicational concept.
I distinguish between a lexical item and the explication of its meaning
that articulates what the individual would give, under some reflection, as
his understanding of the word. Call the former “the word” and the latter
“the entry for the word”. I also distinguish between the concept
associated with the word and the concept(s) associated with the entry.
Call the former “the concept” and the latter “the conceptual explication”.
Finally, I distinguish between a type of meaning associated with the
word, “translational meaning”, and the meaning associated with its entry,
“explicational meaning”. For our purposes, the explicational meaning is
the semantical analogue of the conceptual explication. (1989, 181)
It is important to note that whilst the explicational meaning of a term can be
communal, in the sense of an explication that has been agreed upon and
codified, perhaps as a dictionary definition, it can also be idiolectical, in the
3
8
The dual-content approach has been developed most famously by Block (1986), Fodor (1987)
and Loar (1988) but it was present at the very beginning in Putnam (1975).
sense of the individual’s personal explication of the term: the entry in that
individual’s personal dictionary, so to speak. Neither kind of explicational
meaning is identical to the translational meaning, which can persist through
changes in both idiolectical and communal explications of a concept.
Associated with the two kinds of meaning are two kinds of understanding:
[O]ne must distinguish the sort of understanding of a word in being able
to use it to express a concept or translational meaning from the sort of
understanding that is involved in being able to give a correct and
knowledgeable explication of it. One may think with a concept even
though one has incompletely mastered it, in the sense that one
associates a mistaken conception (or conceptual explication) with it.
(1993, 316f)
Burge’s distinction clearly provides us with an account that accepts and
explains the intuitions behind Consumerism and FPD, without having to
conclude that normal practice is contradictory. The move from the principle
that a rational subject will not sincerely and knowingly assent to contradictory
beliefs, to FPD relies on the Cartesian idea that a subject is introspectively
aware of whether her beliefs contradict. On Burge’s distinction, what is clear is
that a subject will certainly be aware if her beliefs contradict in their
explicational concepts (personal dictionary definition), but may not be aware if
they contradict in their translational concepts. We can now say that intuition
supports FPD concerning explicational concepts, and not translational
concepts.
FPD-E: If a rational subject sincerely and on reflection assents to “a is F”
and “b is not F” then they express different explicational concepts with
“a” and “b”.
Turning to Consumerism, we can use Burge’s distinction here as well to
formulate two principles where previously we have talked of one. It is plausible
to say that intuitions support the principle concerning translational concepts,
but certainly not explicational concepts. A subject does not, through their
ability to be a normal, competent member of a linguistic community, get to be
attributed the standard explicational understanding, dictionary definition, of a
term4. Conceptual explications vary greatly from speaker to speaker, and
where there is a codified and agreed communal explication of a term, it may
not be known by many or even most of the competent speakers of the
language, who are nevertheless perfectly able to use that term in the standard
way to express the standard linguistic concept. However, Consumerism-T is
plausibly supported by intuition. We do, as a fact of normal practice, ascribe
standard linguistic concepts to competent speakers on the basis of their
sincere utterance or assent.
Consumerism-T: if a subject is minimally competent and deferential with
an expression then they express the standard translational concept with
that expression.
4
If we don’t make the optional change and remove talk of experts, it is even clearer that DP-E is
false. I do not get an expert’s understanding of “arthritis” by becoming competent enough to use
it, and be ascribed beliefs involving it.
9
So our natural solution is that if we correctly distinguish between the two
aspects of meaning that speakers can express with terms, the contradiction
that appeared to exist between our intuitions evaporates. Whilst the
undifferentiated principles Consumerism and FPD do seem to contradict, it
can be shown that intuition does not support them. Rather, it supports,
Consumerism-T and FPD-E, which do not contradict. To reiterate: the
consumerist principle that is a part of normal practice concerns the
translational meaning of terms, the linguistic concepts they express; and the
version of The Fregean Principle of Difference that is a part of normal practice
is not a test for difference in translational concept expressed, but for
difference in explicational concept, conceptual explication.
Now we can apply this account of the intuitions to Segal’s scenario that
we considered in section 1: the fact that Alf assents to “arthritis is not an
inflammation of the joints” may reveal that Alf associates different explications
with the translational concepts expressed by the expressions “arthritis” and
“inflammation of the joint”. But that does not mean that the translational
linguistic concepts themselves differ – that is fixed, by his linguistic
competence. Equally, Alf associates the same linguistic concept with the term
“arthritis” as the experts, as they both belong, as functioning members, to the
same linguistic community. But of course Alf does not share the same
explicational understanding of the term as the experts. FPD-E and
Consumerism-T adequately account for the situation described by Segal,
without any threat of contradictory intuitions.
Clearly this is only a very rough gesture at a satisfactory account, and
(as will be discussed later) its plausibility will depend on how the detail is filled
in, but the bare outline is suggestive of a viable, and natural, solution.
5.
Problems for the social externalist
The account given above – which I shall call the “Modification Strategy” –
seems to be a natural and intuitive method of dealing with our apparently
contradictory intuitions, but such dual-content accounts have normally been
proposed by those with internalist inclinations. The question concerns the
issue of cognitive content. All the parties to the debate are concerned to
describe the nature of that property of belief which is relevant to and operative
in normal everyday folk-psychological explanation. An externalist is an
externalist precisely because she believes that cognitive content is
individuated externalistically. And an internalist is such because she believes
that cognitive content is individuated internalistically. Given that fact, the issue
that arises most prominently from the Modification Strategy is this: which
aspect of belief – the socially individuated translational concept relevant to
Consumerism, or the introspectively accessible explicational concept relevant
to the Fregean Principle of Difference – is relevant to cognitive content?5
5
10
In posing this question I am rejecting the possibility that both explicational and
translational concepts can contribute to cognitive content. Normal practice seems to
assume a single notion of psychologically operative content, and we have seen that
explicational and translational concepts pull in different directions in certain situations. It
seems, and I will assume, that we can therefore countenance only one of the aspects
Social externalists place great emphasis on the role of their famous
thought-experiments, and the intuitions that they inspire. The idea is that two
intrinsically identical individuals, Bert and Twin-Bert, can be in different mental
states, purely because of a difference in their linguistic communities. Bert, in
our community, believes that you can get arthritis in your thigh. Twin-Bert, in a
different community, where “arthritis” is used for a condition that can occur in
joints and muscles, does not believe that you can get arthritis in your thigh –
despite saying that “I have arthritis in my thigh”, and being in all intrinsic
respects identical to Bert. This thought-experiment supports the claim that the
twins differ in the translational concepts they express with their terms. But it is
written into the situation that both twins associate the same conception with
the term “arthritis” – their personal dictionary definitions would be the same.
So the thought-experiment so heavily relied upon by social externalists
supports externalism for translational concepts, and not for explicational
concepts. The twins share explicational concepts, and differ in translational
concepts, so in so far as the thought-experiments support social externalism
about the respects in which the twins differ, the thought-experiments support
social externalism about translational content. So in order to use the those
thought-experiments in such a way, the social externalist needs to show that
translational concepts compose cognitive content. There are however
difficulties with this requirement.
It is often held that the idiosyncratic, subjective component of beliefs is
more suitable for the purposes of psychological explanation. There are a
variety of ways to pursue this claim, and I shall adopt only one. I shall pursue
the argument using the issue of rationality6.
The first step in the argument is to claim that issues of rationality are
central to folk psychological explanation. We predict people’s behaviour under
the assumption that they are rational, and make sense of their behaviour in
the light of their rationality. If it turns out that someone is acting irrationally, we
seem to be barred from being able to engage in folk psychology concerning
them. Rationality is central to folk psychological explanation.
The next step in the argument is to claim that what matters to
considerations of rationality is how a subject conceives of the world. It is
rational for person A to do action X because of the way person A sees the
world, their subjective view on the world – the way the world actually is seems
to take a back seat. To use a concrete example: where would it be rational for
Bert and Twin-Bert to apply an ointment labelled “arthritis cream”? Given that
they both conceive of “arthritis” as being an ailment in their thighs, to be
rational they both need to apply the cream to their thighs. This is despite the
fact that in Bert’s (our) world, arthritis just cannot occur in muscles, only joints.
That last fact seems irrelevant to the question of what it is rational for Bert to
do.
of belief being fundamentally contributive, in a substantive sense, to that property of
belief which normal practice takes to be operative in folk psychological explanation.
6
There are many complexities in the subject of rationality which are not directly relevant to this
argument. The issue of rationality is just used here as an intuitive example of a case in which the
subjective quality of belief is relevant to psychological explanation. The question of how
rationality as a whole relates to externalism as a whole is a subject in itself.
11
To use another example. Alf says “arthritis is not the same thing as an
inflammation of the joint”. Alf is, by stipulation, rational. Arthritis is the same
thing as an inflammation of the joint, so it can’t be that, the way the world is,
that renders his utterance rational. What renders his utterance rational is the
fact that the conceptions he associates with “arthritis” and “inflammation of the
joint” differ.
If we add the agreed premise that it is explicational concepts that
capture a subject’s conception of the meanings of her words, we can see that
this argument, in showing that what is relevant to questions of rationality is the
subjective aspect of beliefs, shows that it is explicational concepts that are
central to questions of rationality. Further evidence for this claim is that
rationality is explicitly referred to in our formulation of the Fregean Principle of
Difference, and so the two are deeply linked.
So, explicational concepts (and not translational concepts) are relevant
to rationality. And we said above that rationality is central to folk psychology.
So we have to conclude that it is therefore explicational concepts that are
central to folk psychology, and cognitive content. Translational concepts, on
the other hand, seem to have therefore no central role in folk psychology.
So while the thought-experiments support social externalism about
translational concepts, they do not seem to compose cognitive content. The
conjunction of a) the natural way of dealing with the apparent contradictions in
our intuitions (the Modification Strategy) and b) the argument just given,
seems to render the traditional thought-experiments incapable of supporting
social externalism. A social externalist who adopts the Modification Strategy
“is effectively conceding that [Consumerism] is inadequate to account for
[Alf’s] behaviour….[T]he difference between the taxonomy of belief contents
offered by [Consumerism] and the taxonomy for explaining actions is being
recognised” (Patterson 1990, 321).
6.
Social externalist response
What kinds of responses are open to the social externalist at this point? She
could give up the dual-content approach and try and explain away the
intuitions supporting the Fregean Principle of Difference. Such intuitions are
very powerful, and the task seems daunting. Alternatively, the social
externalist could attempt to retain the Modification Strategy, and defuse the
argument just given, and claim that rationality is not as central to folk
psychology as has been claimed. This could be done by showing that whilst
our conception of the world is what matters when we explain our own
behaviour, when we explain the behaviour of others we do not focus so much
on their subjective viewpoint. We take into account many things which the
subject may not be aware of, or are not aware of in the same way as us. In
short, we may take into account the context in which the subject is embedded,
and explain such behaviour without explicit concern for the rationality of the
subject. We may, for instance, care about the suitability of the behaviour. This
is a large issue, but it seems a well-supported claim, that what we care, in
some sense, most about when we explain and predict people’s utterances
and physical behaviour is the state of their conception of the world and the
meanings of the words they use, which I am tying to the issue of rationality.
12
The response I want to formulate is a different one, and concerns the
possibility of accepting the argument concerning rationality, but spelling out
the detail of the Modification Strategy in such a way as to avoid the apparent
problems for social externalism.
The problem, recall, is that it is the kind of content that Alf and the
experts do not share – the explicational concept – that seems to be relevant
to folk psychology. Alf and the experts associate different explicational
concepts with the term “arthritis”: Alf believes that you can get arthritis in
muscles, while experts do not, and they have many further technical beliefs
about arthritis that Alf does not share. What appears to be the case here,
however, is that it is quite easy to specify the differences in their explicational
concepts, using perfectly standard words with perfectly standard meanings.
The idiosyncrasy seems not to go beyond the surface.
[This sort of case] demonstrates the strength of linguistic content, not its
commonly supposed weakness: even in those bizarre cases the appeal
to linguistic contents and other non-supervenient facts about history and
belief acquisition is perfectly sufficient to explain what’s going on.
(Frances, 1999, 63)
This sort of view is suggested by Burge in his own comments on similar
puzzle cases:
You ask me what Al would think. It would be misleading for me to reply
that Al would think that you do not have a contract, if I know that Al
thinks a contract must be based on a formal document….In such cases,
it is incumbent on us to cite the subject’s eccentricity: “(He would think
that you do not have a contract, but then) he thinks that there is no such
thing as a verbally based contract.” (1979, 91)
In describing Alf’s explicational concept of arthritis, we consider it perfectly
acceptable to ascribe certain standard concepts, without any hint that they
themselves are idiosyncratically understood, or “private” in any sense – such
as “he thinks one can have arthritis in a muscle”. Public concepts attributed to
competent language-users according to principles sensitive to linguistic
conventions seem perfectly adequate to explicate the most bizarre and
idiosyncratic understanding of a term. On this view, a subject’s idiosyncratic
explicational concept can be specified using words expressing perfectly
standard public linguistic concepts, i.e. translational concepts. What this
response does concede is that translational concepts as ascribed by
Consumerism, in the form given above, are often not sufficient for
psychological explanation. We may need to ascribe translational concepts in a
different way in order to explain how the subject idiosyncratically groups them
together in their conception of the meaning of a term:
Of course, relying on simple disquotation will get it wrong in such cases.
Being more complicated than the normal case, they require different,
and often more complex, that-clauses if they are to be described
accurately. (Biro 1992, 288)
We can cast this response in terms of the privacy and publicity of concepts:
the externalist can concede that explicational concepts are private in the
sense that they may indeed not be shared by anyone else, but they are public
13
in that they can be, and often are, clearly communicated. So whilst
explicational concepts are paramount in folk psychological explanation, their
use is underpinned by that of translational concepts, which are, as we have
seen, individuated according to social externalism.
We can draw out the core issue by reconsidering the original thoughtexperiment, with the twins Bert and Twin-Bert. Their shared explicational
concepts seem to be doing the work in any folk psychological explanation we
may want to give of them. And we can specify that shared explicational
concept. So whilst we can’t say that they both believe that one can have
arthritis in their thigh – as Twin-Bert has no beliefs about arthritis – we can say
that they both believe that one can have “arthritis” in the thigh, that old people
get it, that their Auntie Doris has it, and all the other things that warrant the
claim that they share the same explicational concept. That is, we are inclined
to say that they have some important property of belief in common because
we are told that they both think that one can have “arthritis” in the thigh etc.
But what is now apparent is that there is an assumption that they don’t
really inhabit different linguistic communities. For the purposes of all terms
except “arthritis”, they inhabit the very same linguistic community, so we can
specify, without difficulty, what it is that they share. What the social externalist
may be able to claim is that if the difference in their linguistic communities
were expanded to include certain other terms along with “arthritis”, then the
impulse to say that they shared any important belief content in common would
recede. Specifically, if their linguistic communities used the terms “thigh”,
“old”, “Auntie” in a systematically different way, consonant with their intrinsic
identity, then we may lose the ability to claim that they share any explicational
concept. We would have to say that they both believe that one can get
“arthritis” in the “thigh”, that “old” people get it, that their “Auntie” Doris has it.
And this seems to say next to nothing about what they believe, and be of next
to no use in any folk psychological explanation. When we lack translational
concepts with which to specify a subject’s explicational concept, the pull to
claim that such an explicational concept actually exists is not there. The drive
to say that Bert and Twin-Bert share a psychological important property of
belief is a product not of their intrinsic similarities, but their shared linguistic
environment and their responsibility to shared linguistic norms.
This response claims that the thought-experiments seem only to support
social externalism concerning translational concepts because they are not
really cases of subject being in different linguistic communities. If we expand
the thought-experiment in the way described, it becomes apparent that
explicational concepts are socially externalistic, in that they are composed of
the translational concepts that have been socially externalistic all along. That
is, explicational concepts themselves are sensitive to changes in linguistic
environment.
Conclusion
If the contradictory intuitions raised by the social externalist rebuttal of Segal
do force us to adopt a tactic such as the Modification Strategy, what is clear is
that there is a lot to be done in order to make a social externalist approach to
14
folk psychological explanation plausible. What seems to be the case is that a
simplistic version of social externalism such as Consumerism (or the
Disquotational Principle) is inadequate to capture the complex relationships
between a subject and their environment. Those externalist efforts focussed
on such uncomplicated principles may be misguided, and as we have just
seen, success for the externalist may lie in distancing themselves from such
principles and developing detailed externalist accounts of a subject’s
idiosyncratic conception of the world.
References
Biro, J. (1992), “In Defence of Social Content”, Philosophical Studies 67: 277293.
Block, N. (1986), “Advertisement for a Semantics for Psychology”, Midwest
Studies in Philosophy 10, French et al eds.
Burge, T. (1978), “Belief and Synonymy”, Journal of Philosophy 75(3): 119138.
Burge, T. (1979), “Individualism and the Mental”, in P. French, T. Uehling, and
H.
Wettstein, eds., Studies in Epistemology, Midwest Studies in Philosophy 4
(Minneapolis, University of Minnesota Press): 73-121.
Burge, T. (1986), “Intentional Norms and the Foundation of Mind”, Journal
of Philosophy 83(12): 697-720.
Burge, T. (1989), “Wherein is Language Social”, in A. George, ed., Reflections
on Chomsky (Oxford, Blackwell): 175-191.
Burge, T. (1993), “Concepts, Definitions, and Meaning”, Metaphilosophy
24(4): 309-325.
Fodor, J. (1987), Psychosemantics (London, MIT).
Frances, B. (1999), “On the Explanatory Deficiencies of Linguistic Content”,
Philosophical Studies 93: 45-75.
Kaplan, D. (1989), “Afterthoughts”, in J. Almog, J. Perry, and H. Wettstein,
eds., Themes From Kaplan (Oxford, OUP): 565-614.
Kripke, S. (1979), “A Puzzle About Belief”, in A. Margalit, ed., Meaning and
Use (Dordrecht, Reidel). Reprinted in N. Salmon, and S. Soames, eds.,
(1988) Propositions and Attitudes (Oxford, OUP): 102-148.
Loar, B. (1988), “Social Content and Psychological Content”, in R. Grimm, and
D.
Merrill, eds., Contents of Thought (Tucson, University of Arizona Press).
Reprinted in D. Rosenthal, ed., (1991) The Nature of Mind (Oxford,
OUP): 568-575.
Owens, J. (1989), “Contradictory Belief and Cognitive Access”, in P. French,
T. Uehling, and H. Wettstein, eds., Contemporary Perspective in the
15
Philosophy of Language 2, Midwest Studies in Philosophy 14
(Minneapolis, University of Minnesota Press): 289-316.
Owens, J. (1992), “Cognitive Access and Semantic Puzzles”, in Anderson,
and Owens, eds., Propositional Attitudes (CSLI Press): 147-173.
Patterson, S. (1990), “The Explanatory Role of Belief Ascriptions”,
Philosophical Studies 59: 313-332.
Segal, G. (2000), A Slim Book About Narrow Content (London, MIT Press).
Segal, G. (2003), “Ignorance of Meaning”, in A. Barber, ed., Epistemology of
Language (Oxford, OUP).
16
Transcendental Arguments in the Philosophy of Mind
Mathilde Byskov Jakobsen
Department of Philosophy, University College London
m.jakobsen@ucl.ac.uk
Abstract
In this paper I defend the use of transcendental arguments within the
philosophy of mind. I argue that we can use transcendental arguments to
establish certain necessary features and capacities of the mind, given that it is
the mind of a certain kind of experience. This means that as well as
formulating belief-directed transcendental arguments, we can also formulate
truth-directed transcendental arguments as long as they are directed at
certain features of the mind. I present an argument for the unity of
consciousness to illustrate this form of transcendental argument. I argue that
this is a truth-directed transcendental argument which is not subject to
Stroud’s objection to the use of such arguments outside of an idealist
framework.
1.
Introduction
My aim in this paper is to defend the use of transcendental arguments in the
philosophy of mind. I will identify a certain kind of belief-directed
transcendental arguments that can be of some limited use in the philosophy of
mind. This use of transcendental arguments fits with the view that all
transcendental arguments can ever hope to reach are conclusions about
beliefs. However, the main claim which I will defend in this paper is that a
certain kind of truth-directed transcendental argument can also be of use in
the philosophy of mind. I will defend this claim primarily by considering the
methodology of transcendental arguments directed at the mind, but also by
developing an example of a self-directed transcendental argument. I call this
type of argument a “self-directed transcendental argument” following Quassim
Cassam, as it is a name which reflects its purported subject matter. My paper
has three main parts. First, I will consider the methodology of transcendental
arguments in general and present Barry Stroud’s well-known objection to
them. Second, I will present two kinds of self-directed transcendental
arguments of use in the philosophy of mind, which are not subject to this
objection. The first of these is a kind of belief-directed transcendental
argument. The second kind is a kind of truth-directed transcendental
argument. The possibility of formulating truth-directed transcendental
argument of this kind is exclusive to the philosophy of mind, and it is
dependent on an assumption about the nature of mind which is broadly
functionalist. In the third part of the paper I will further evaluate this strategy by
considering an example of a self-directed transcendental argument. This
argument will be developed from some of Kant’s claims about transcendental
apperception in the ‘Transcendental Deduction’. I conclude that this
methodology can identify significant constraints on the form of the mental
17
operations that our minds must implement, given that our experience has the
character that it does, and so tell us something about the mind – or the self.
I
2.
What are transcendental arguments?
Transcendental arguments argue from premises about experience to a
conclusion about reality. Typically, they start with a premise which states that
we have experience of a certain kind. The second premise determines that
this form of experience is conditional upon some unobvious and
philosophically interesting claim about reality. Which we can then conclude
must hold. In one traditional use of such arguments, the aim of the argument
is to refute the philosophical sceptic. In this traditional use the first premise is
selected in order to be immune from Cartesian scepticism about the external
world. This premise describes experience in a way which cannot be doubted
from the first person perspective. However, this aspect is optional:
transcendental arguments are not, primarily, intended to refute the
philosophical sceptic even if they have, traditionally, been used in this way.
Rather they are concerned essentially with identifying necessary conditions
for experience, hence the force of the description “transcendental”. This
means that instead of having the following form:
Necessarily P
Necessarily, (if P then Q)
Therefore, necessarily Q
The transcendental arguments I will discuss have this logical form:
P
Necessarily, (if P then Q)
Therefore, Q
If the premise we start with is only contingent, - if it is taken as a matter of fact
that we have a certain kind of experience, then the conclusion reached is not
necessary. And if we cannot establish the first premise with certainty, then the
conclusion reached is only conditional. This means that the argument cannot
be used to refute philosophical scepticism. On the other hand it is not
dependent on the claim that the experience we have is in some sense
necessary.1
An example of a transcendental argument can be found in Kant’s
refutation of idealism (B275):
I have experience of myself as determined in time;
Time determination is dependent on the experience of something
persistent, and this cannot be merely a representation in me;
1
18
Harrison (1982) p. 216.
Therefore, there must be something persistent outside me of which I
have direct consciousness by which I can judge the passing of time.
3.
Stroud’s Objection
In a well-known article from 1968 Stroud objected to this entire class of
arguments. His main objection is that transcendental arguments cannot draw
conclusions about reality, because it is always open to an objector to insist
that it is enough that the world appears to us a certain way or that we believe
it to be a certain way in order for us to have experience – it is therefore not
necessary that the world is in fact in this way. We can apply this objection to
Kant’s argument above. It draws a conclusion about reality – that there is in
fact something persisting outside of me of which I can be conscious, but, the
objection goes, it is in fact enough if it just appears to me as if there is
something persistent outside of me and that I believe that it is independent of
me, in order for me to have experience of myself as determined in time.2 If this
is true, then the transcendental argument has reach a conclusion about some
belief or experience that a thinker must hold, given that the thinker holds other
beliefs or have other kinds of experiences. It then does not in fact argue from
a premise about experience to a conclusion about reality. TA’s are therefore
only belief-directed, rather than truth-directed and do not establish objective
conclusions about how reality must be.
This objection, however, does not hold against the use of transcendental
arguments within the framework of transcendental idealism, as Stroud has
also made clear.3 Such a framework secures the bridge between experience
and reality, because reality, as phenomena, is dependent on our experience
of it. Therefore, in drawing conclusions about necessary features of our
experience using transcendental arguments one is drawing conclusions about
phenomenal reality. As Kant puts it “The conditions of the possibility of
experience in general are likewise conditions of the possibility of the objects of
experience, and (…)[for this reason they have objective validity in a synthetic
a priori judgement” (B197). This means that within the framework of
transcendental idealism we can formulate truth-directed transcendental
arguments.
II
4.
Psychological
belief-directed,
self-directed
transcendental
arguments
We can concede to Stroud that transcendental arguments must be interpreted
as belief-directed if they are to be valid outside the framework of
transcendental idealism, and still preserve the usefulness of such arguments.
We can do that by re-deploying these arguments, to be not about the world,
2
Stroud (1968).
3
Stroud (1999) p. 160.
19
but about how we must take the world to be – not about facts but about
beliefs.
So far I have discussed transcendental arguments as directed at either
the world or a belief about the world. However, we can also direct
transcendental arguments at the mind or the self or at a belief about the mind
or the self. I will call these arguments self-directed arguments, following
Cassam.4 Self-directed transcendental arguments argue from premises about
experience to a conclusion about a belief or a fact about a feature or a
capacity of the mind or the self which is necessary for the experience in
question.
We can produce the following grid of the two kinds of transcendental
arguments so far identified:
World
or
Self/Mind
1: Truths
2a: Beliefs
2b: Experiences
The lesson learned from Stroud is that in a non-idealist framework we
can only formulate transcendental arguments of the second, belief-directed,
kind.
Cassam’s argument in Self and World is an example of a belief-directed,
self-directed transcendental argument. The claim of the transcendental
argument is that it is necessary that a subject of self-conscious experience
has a psychological belief with a certain content: oneself qua subject as a
physical object. Cassam develops both a belief and an experience version of
this argument. He develop the experience version of the argument because
the belief-version is subject to the objection that people who believe that they
are Cartesian souls are still able to ascribe their representations to
themselves is solved by saying that they in fact still have an experience or an
intuition of themselves as physical objects, however confused they are in their
beliefs.
Belief-directed transcendental arguments clearly have a use in the
philosophy of mind. Arguments of this type can show the necessity of a
certain psychological belief that a subject must have in order to have
experience. I call these kinds of transcendental arguments ‘psychological
belief-directed arguments’ for this reason. Such arguments can show that a
subject, who has experience of a certain kind, must have some actual belief
with a specific content. The belief is psychologically necessary for the
possibility of experience. However, all such arguments are subject to what
Cassam calls the ‘misconception’ objection (Self and World). The
misconception objection turns on the fact that if a transcendental argument
can show that some belief is necessary for experience, then it should not be
possible to have a misconception about the subject matter of this belief. If a
belief-directed transcendental argument for instance shows that you cannot
have experience of yourself in time without believing that there are objects
4
20
Cassam (1999).
outside of you, then the existence of sincere idealists, who experience
themselves as determined in time, is impossible.
Psychological belief-directed, self-directed transcendental arguments
can only be used to show that some belief, which we in fact all hold, is
necessary. This obviously does not make such argument useless: if
successful, the transcendental argument will tell us that it is not a contingent
matter that we all hold the same belief. However, it does limit the subject area
of such arguments radically.
However, the use of transcendental arguments in the philosophy of mind
is not limited to establishing the status of certain beliefs that we all hold.
Rather, philosophy of mind is an area where we can formulate truth-directed
transcendental arguments, or so I will argue. In order to argue for this claim, I
will first identify an alternative way in which transcendental arguments can be
belief-directed and met Stroud’s objection. Second, I will argue that in the area
of the philosophy of mind some such belief-directed, self-directed
transcendental arguments are in fact also truth-directed transcendental
argument. If this is right, then transcendental arguments have the potential to
draw conclusions about facts about the mind or the self, not just about beliefs
about the mind or the self.
5.
Theoretical belief-directed, self-directed transcendental arguments
We introduced psychological belief-directed transcendental argument as a
reaction to Stroud’s objection to truth-directed transcendental arguments.
Psychological belief-directed transcendental arguments met Stroud’s
objection by making their aim less ambitious. Psychological belief-directed
arguments aim to draw the conclusion that a subject of experience must have
a certain belief, not that a certain fact must be true of the world or the subject.
However, that is not the only way in which we can make our arguments
less ambitious and so avoid the problem Stroud points out. We can re-deploy
transcendental arguments to
reach conclusions about our conceptual
scheme, instead of about the world. This solution can be used on Strawson’s
arguments in Individuals. On this view transcendental arguments can aim to
show us that a certain belief is part of our conceptual scheme. This
designates a different way in which transcendental arguments can be beliefdirected than the way discussed above. Transcendental arguments can aim to
show conceptual connections between the way in which we, as theorists, think
of experience, and the way in which we, given that, have to use related
concepts. This importantly does not place any psychological constraints on
the beliefs of individual people. I call this second sense of a belief-directed
transcendental argument, a theoretical belief-directed argument, because it
shows the theoretical necessity of a belief, rather than the psychological
necessity of the belief. A transcendental argument could for instance attempt
to establish that given our conceptual scheme it is necessary that we take
objects to be in space and time. The conclusion, that objects are in space and
time, is a belief rather than a fact or truths about the world. However, the
argument does not establish that single individuals must hold this belief to
have experience. This distinguishes theoretical belief-directed transcendental
arguments from psychological belief-directed transcend-dental arguments.
21
Cassam claims that all belief-directed transcendental arguments are
subject to the misconception objection [Self and World]. However, in fact the
misconception objection is only effective against psychological belief-directed
arguments. Sincere idealists are not a problem for the transcendental
philosopher who formulates theoretical belief-directed arguments. She is
committed only to saying that a certain belief is (theoretically) required given
our conceptual scheme not that it is psychologically required, and so allows
that people can be confused and still have experience. A theoretical beliefdirected version of Cassam’s argument for the necessity of having the belief
that one is a physical object to have self-conscious experience would not be
subject to the misconception objection. It would just show that we have good
philosophical reasons given our concepts of experience and subjects to think
that subjects have to be physical objects, without demanding that the subject
must actually believe itself to be one. This argument meets Stroud’s objection.
It does that because it allows that, though it is part of our conceptual scheme
that we take ourselves to be physical objects, we cannot prove that we really
are physical objects. It allows for the possibility that the world may in fact be
very different from the way we have to take it to be.
We can add this last type of transcendental argument to the grid from
before, so that it now looks like this:
1: Truths
World
or
Self/Mind
2a: Beliefs
2b: Experiences
3: Beliefs
} psychological
} theoretical
Theoretical belief-directed transcendental arguments are not very
interesting, when directed at the world in a non-idealist framework. Outside of
transcendental idealism conditions of cognition are not also conditions of
objects. The way in which we have to think of things does not tell us anything
about how they are. In the case of world-directed transcendental arguments,
we can for instance attempt to show that the belief in causation plays a central
role in our conceptual scheme, so that in a reflection on the possibility of
experience, we have to think of the world as causally ordered. However, the
question will always arise whether independent reality actually corresponds to
this necessary way of conceiving of it. The fact that our conceptual scheme
demands that we make sense of the world only as causally ordered does not
tell us what we want to know, namely whether the world is in fact causally
ordered or not.
6.
The nature of mind
However, my suggestion is that in the case of at least some of the
transcendental arguments we can direct at the mind, we do not find the same
gap between how we must think of things and how they really are. Say that
we have established that we must necessarily interpret or conceive of the
mind as having certain features and characteristics in order to understand it
as a mind which has experience of a certain kind. Now, we cannot allow that it
22
may in fact not really have these features at all – whilst still being interpretable
as a mind of this kind of experience.
In cases where our argument shows that we must necessarily think of
the mind as realising certain functions or features related to functions there is
no sense in driving a wedge in between the claim that the mind must be
interpreted or conceived as having these features and capacities, and the
claim that the mind really realises these features and capacities. The picture
of the mind required to support this view is broadly functionalist in that a mind
is defined by its functions.5 If something is defined by its function, and it
performs these functions, then there is no question as to whether its real
nature fits our functional picture of it.
This assumption about the mind opens up the possibility that if we can
formulate transcendental arguments about features of the mind necessary for
it to perform its functions, then we can consider them as identifying aspects of
the mind and not just of our conception of the mind.
This assumption about the mind must not be confused with the claim
that the mind is transparent to itself either in the sense that we know infallibly
the features of our minds or the content of our mental states by direct
introspection. It is not a claim about people’s psychological relation to their
empirical minds.
It also does not suggest that we can find out everything there is to know
about the mind by the use of a transcendental reflection on the possibility of
experience. First, there are many empirical facts about the mind which require
empirical methods. Second, transcendental philosophy cannot tell us how a
certain necessary feature or capacity is realised in the mind – if there is more
than one way in which this could be done. A transcendental philosophy of
mind uses self-directed transcendental arguments to ascribe features to the
mind (necessary conditionally dependent on that the mind in question is a
mind of a certain kind of experience). However, while it tells us that the mind
must realise these features, it does not tell us how it realises them. Just like
other kinds of functionalism, transcendental philosophy of mind is ontologically
uncommitted. It can tell us nothing about either the substrata of mind or about
the actual realisation of the very general constraints we can place on the
mind.
This suggests that the philosophy of mind allows for transcendental
arguments to be used to establish facts about the mind and not just beliefs
about the mind.
III
7.
An example
To take a concrete example of this strategy, consider Kant’s account of
transcendental apperception in the B-edition of the ‘Transcendental
5
It is only ‘broadly’ functionalist as it is not committed to any of the more specific
requirements of contemporary functionalism.
23
Deduction’. A self-directed transcendental argument can be developed out of
some of these considerations. The starting point of the B-Deduction is the
observation that experience is made up of complex representations, where
both the individual elements of a complex representation and different
complex representations are synthesised together to form one experience.
The next step is to say what this requires in the subject of such experience.
The answer for Kant is the unity of transcendental apperception:
It must be possible for the ‘I think’ to accompany all my representations;
because otherwise something would be represented in me which could
not be thought at all and that is equivalent to saying that the
representation would be impossible, or at least would be nothing to me.
(B132).
This quote is saying that all conscious parts of one experience (all
representation that are not noting to me) must be self-ascribable to one ‘I’.
The idea that a bundle of representations are all in some sense mine
corresponds to one understanding of the unity of consciousness in
contemporary philosophy of mind. We have to distinguish this understanding
of the unity of consciousness from one that refers to a feature of our
experiences themselves. The term ‘the unity of consciousness’ can be used to
refer to a phenomenal feature of experience: the fact that our experiences are
unified in the sense that it feels different to experience a together with b, than
it does to experience a alone (and that it is the experience of a, which is
different).6 Unity of consciousness, as I am interested in it, is not a claim about
the feel of experience. If it was the argument would be trivial as the claim that
experience is unified (at least partly) is already involved in the claim that it
contains complex representations. Saying that a bundle of conscious states
must all be ascribable to the same ‘I’ , saying that a bundle of representations
all in some sense belong to me, is not a claim about experience but about the
mind. So the possibility of complex representations requires that the parts of
this representation are in some sense all mine – which is equivalent to the
claim that they belong to one unity of consciousness. I will suggest three ways
in which we can understand the unity of consciousness understood in this
way.
One way of understanding what it means to be self-ascribable to one ‘I’
is via a causal relation to a body – a bundle of experiences are part of a unity
of consciousness if they are all causally dependent on the same body. We
can reject this view because there are no good reasons to think that there is a
one-one relation between a body and a unity of consciousness – and split
brain and multiple personality disorder cases in fact give us reasons to think
that this is not necessarily the case. The conceivability of a unity of
consciousness supported by more than one body – perhaps by a succession
of bodies as in Kant’s billiard ball example also supports this point. The fact
6
24
Dainton and Chalmers and Bayne take the unity of consciousness to express a
phenomenal quality about our experiences themselves at a time. For Dainton the unity
of consciousness is a phenomenal characteristic that lies in the co-conscious
experiences themselves, for Chalmers and Bayne it is the existence of a state, whose
phenomenal character subsumes the phenomenal characters of all the states the
subject is in at that time.
that a bundle of representations all belong to a body in being causally
dependent on it does not mean that they can all be self-ascribed together.
And the possibility that a bundle of representations are causally dependent on
two bodies does not rule out that they could be self-ascribed to one subject
together. And it is this latter sense of belonging we need to make complex
representations possible. This shows that just because something is true of
the subject it is not necessarily true for the subject - it is not enough that it is
true that the parts of a representation are mine if I don’t take them to be mine.
Another option is to understand the unity of consciousness as a claim
about actual self-ascription. Under this suggestion a number of conscious
states are part of a unity of consciousness if there exist in the subject a state
of the form ‘I think that…’ which subsumes all of them. However, the problem
with basing unity of consciousness on actual self-ascription is that it limits it to
include only states where such a higher order state exists, which seems too
restrictive. It is after all supposed to be necessary for the possibility of
complex representations.
We therefore have the further option of basing the unity of
consciousness on possible self-ascription, which avoids this problem.
According to this suggestion a bundle of experiences are parts of a unity of
consciousness if they can be self-ascribed together under one ‘I think’. Some
people object to the use of dispositions or possibilities in an account of the
unity of consciousness because it is hard to make the claim substantial. In
situations where a subject at a time has a whole range of experiences, only
some of which are ever actually taken together, how do we know that it could
actually have done so to all of them? However, the solution cannot be to avoid
talk of possibilities in favour of actualities. Accounting for the unity of
consciousness in terms of a disposition or a capacity for self-ascription is still
the best option, in my view, because it captures the sense of experiences
belong to me, that we need, without being unnecessarily restrictive.
If experience contains complex representations, then the subject of
experience realises a unity of consciousness – that is, the parts that make up
the complex representations can all be self-ascribe under one ‘I think’. Kant
illustrates the necessity of the unity of consciousness with the example of the
consciousness of a verse of a song – in order for there to be a consciousness
of a verse, it is necessary that all the parts of the verse is taken by the subject
as all its experiences and as belonging together. If a conscious complex
representation is to count as such, then the subject must be able to take its
part as a whole and self-ascribe it as one. There is no need for the subject to
actually so self-ascribe it, which shows that it is the possibility not the actuality
of self-ascription which is important. So if experience contains complex
representations, then the unity of consciousness is necessary for experience.
8.
The status of the conclusion
In this argument we go from an observation of the kind of experience we have
to a conclusion about a necessary feature of the mind that has such
experience, which in this case is the unity of consciousness. This conclusion
of the argument is not that the subject must believe in the unity of
consciousness in order to have experience. If the unity of consciousness is
necessary for the possibility of experience containing conscious complex
25
representations, then it is neither necessary nor sufficient that the subject
believes that it realise a unity of consciousness, it must actually do so. The
argument does make a claim our conceptual scheme. The conclusions we
draw about the mind on the basis of self-directed transcendental arguments
link ways in which we think of experience with ways in which we, given this,
must think of the mind. The argument identifies a conceptual connection
between thinking of experience as unified and thinking of the subject of these
experiences as realising a unity of consciousness. Describing this connection
as conceptual does not reduce it to the trivially linguistic. It involves a
reflection on experience, and a reflection on how we must think of the
experiencer given that it is the subject of this kind of experience. However, the
conclusion is not restricted to a claim about how we must think of the mind. In
the case of functionally defined aspects of the mind, if we can show that we
must necessarily think of the these features in a certain way, given that it is
the mind of a certain kind of experience, then this identifies a real aspect of
the mind.
It seems to me plausible to defend this general claim in the particular
case of the argument about the necessity of the unity of consciousness.
This particular use of a self-directed transcendental argument, then, is
not subject to Stroud’s general methodological strictures: It is not restricted to
showing only the necessity of a psychological belief that we all hold and it is
not restricted to showing only something about our general conceptual
scheme – if we take this to mean that it has not also been established as a
feature of reality. The argument is therefore truth-directed.
I conclude that the philosophy of mind allows for the use of
transcendental arguments to determine the features of the mind that are
conditionally necessary for our experience to take the form that it does.
Bibliography
Brook, Andrew: Kant and the Mind. Cambridge University Press 1994.
Cassam, Q: Self and World. Oxford University Press 1997.
Cassam, Q: ‘Self-directed Transcendental Arguments’ in
Transcendental Arguments. Oxford University Press 1999.
Stern,
R:
Chalmers, D. and Bayne, T: ‘What is the Unity of Consciousness’. Here from
Chalmers’ webpage, but published in A. Cleeremans (ed.), The Unity of
Consciousness: Binding, Integration, Dissociation. Oxford University
Press 2003.
Dainton, B: Stream of Consciousness. Routledge 2000.
Gardner, S: Kant and the Critique of Pure Reason. Routledge 1999.
Harrison, R: ‘Transcendental Arguments and Idealism’ in Vesey, G: Idealism,
Past and Present. Cambridge University Press 1982.
Kant, I: Critique of Pure Reason. (Translated by Kemp Smith) The Macmillan
Press 1929 (second edition 1933).
Rosenthal, D: ‘Unity of Consciousness and the Self’ Proceedings of the
Aristotelian Society 2003.
26
Strawson, P. F: Individuals. Routledge 1959.
Stroud, B: ‘Transcendental Arguments’ The Journal of Philosophy vol. LXV,
no. 9. 1968.
Stroud, B: ‘The Goal of Transcendental Arguments’ in
Transcendental Arguments. Oxford University Press 1999.
Stern,
R:
Stroud, B: ‘The Synthetic A Priori in Strawson’s Kantianism’ in Glock, H-J:
Strawson and Kant. Oxford University Press 2003.
27
Conceptualism and The (Supposed) Non-Transitivity of Colour
Indiscriminability
Charles Pelling
Department of Philosophy, University of Reading
Abstract
In the philosophy of perception, the conceptualist asserts, and the nonconceptualist denies, that the representational content of experience is
exclusively conceptual content. In this paper, I show that Delia Graff’s recent
arguments for the traditionally unpopular view that colour indiscriminability is
transitive have important implications for the conceptualism versus nonconceptualism debate.
This is because (1) conceptualism can be true only if we possess
context-dependent demonstrative colour concepts, and (2) only if colour
indiscriminability is transitive can we possess such concepts.
This paper addresses (2). In order to establish its truth, I consider two
accounts of demonstrative colour concept possession, those given by the two
most prominent conceptualists, John McDowell and Bill Brewer. McDowell
and Brewer each propose a condition that a subject must satisfy in order to
possess a demonstrative colour concept. In the bulk of the paper, I am
concerned to establish two things: first, that unless colour indiscriminability is
transitive, neither of the conceptualists’ proposed conditions are satisfiable;
and second, that at least one of these conditions must indeed be satisfied by
any genuine demonstrative colour concept possessor.
1.
Introduction
Delia Graff has recently argued – effectively, I think - for the traditionally
unpopular view that colour indiscriminability is transitive.1 My aim in this paper
is not, however, to defend or even to discuss Graff’s arguments, but rather to
investigate their implications for an issue in the philosophy of perception that
has generated considerable debate in recent years.
The debate is between the conceptualists – such as John McDowell and
Bill Brewer - and the non-conceptualists – such as Christopher Peacocke and
the late Gareth Evans. After explaining briefly what these opposing positions
amount to, I will show that there are good reasons for the conceptualists to
accept Graff’s view that colour indiscriminability is transitive.
2.
Conceptualism and Non-Conceptualism
Conceptualism is the view that the representational content of (normal human
adult) perceptual experience is exclusively conceptual content. Non-
1
In her 2001. Graff’s arguments extend also to other forms of indiscriminability. I will
focus on the issue of colour indiscriminability in order to simplify the discussion.
29
conceptualism is the view that the representational content of perceptual
experience is at least partly non-conceptual content.
As I use the terms ‘conceptual content’ and ‘non-conceptual content’,
insofar as the content of a subject’s experience is conceptual, the experience
has the content it does in virtue of its drawing into operation concepts that the
subject himself possesses at the time of the experience; insofar as the content
of a subject’s experience is non-conceptual, by contrast, the experience has
the content it does in virtue of something else – not in virtue of its drawing into
operation concepts that the subject possesses at the time of the experience.
3.
Conceptualism and Fineness of Grain
An immediate consequence of the conceptualist view is that we must have
concepts for all the various objects, properties, and relations which feature in
our experiences. A traditional worry here is that it seems unlikely, initially at
any rate, that we have a concept for every colour shade (say) that features in
our experience when we are looking at a spectrum. It seems tempting to say
that in such a situation, our colour concepts are not as fine-grained as the
shades that feature in our experience.2
However, the conceptualists insist that this worry rests on the mistaken
assumption that we can possess only those context-independent concepts
expressible by such general words as ‘red’, ‘green’, or ‘terracotta’. According
to the conceptualists, such an assumption overlooks the possibility that while
one is enjoying an experience in which a particular colour shade features, one
can express a concept of exactly this shade by using a phrase which includes
a demonstrative.3 This response, the conceptualists contend, is sufficient to
undermine the worry that stems from the issue of fineness of grain since it
seems that one could have such a demonstrative concept for any shade, or
any other property or relation, which could conceivably feature in an
experience.
4.
Demonstrative Concepts and Recognitional Capacities
But the conceptualist, as McDowell himself notes, must ensure that our
supposed demonstrative concepts really are bona fide concepts.4 McDowell
draws our attention to Wittgenstein’s case in which a subject proclaims ‘I know
how tall I am!’ and lays his hand on top of his head to prove it.5 McDowell
agrees that such a subject would lack a genuine demonstrative concept of his
own height. But given this, what exactly is required for a subject to possess a
demonstrative concept of a particular height (or a particular shade, etc.)?
McDowell’s suggestion is that:
‘We can ensure that what we have in view is genuinely recognizable as
a conceptual capacity if we insist that the very same capacity…can in
2
cf. Evans, 1982, p. 229, and Heck, 2000, pp. 489-90.
3
See McDowell, 1994, pp. 56-7, and Brewer, 1999, pp. 170-4.
4
McDowell, 1994, p. 57.
5
Wittgenstein, 1958, § 279.
30
principle persist beyond the duration of the experience itself…What is in
play here is a recognitional capacity, possibly quite short-lived, that sets
in with the experience.’6
To establish how this proposal is supposed to work, let us return to the case
where a particular colour shade is featuring in a subject’s experience. I think
that McDowell’s idea, then, amounts to this: the subject possesses a concept
of the shade only if he is capable, after the shade no longer features in his
experience, of telling whether or not a shade that features in a new
experience falls under the very same concept.
So for instance, suppose that my experience represents an object as
being a particular shade of blue. According to McDowell’s proposal, if I
possess a concept for this exact shade, I must be capable, after the shade no
longer features in my experience, of telling whether or not something is being
represented to me in a new experience as being this exact shade of blue.
However, not all philosophers have been convinced by McDowell’s
account of our possession of such demonstrative concepts. One common
suggestion is that, as a matter of empirical fact, shades7 just do feature in our
experiences even when we lack the corresponding recognitional capacities
that McDowell envisages: contra McDowell, a shade can feature in my
experience even when I would not be able to tell, after the shade no longer
features in my experience, whether or not a shade featuring in a new
experience is identical to the original shade. Sean Kelly voices something like
this worry when he claims that:
‘there’s nothing in the nature of perception to keep it from being true, that
our capacity to discriminate colours exceeds our capacity to re-identify
the colours discriminated.’8
And if Kelly is right, then either McDowell is wrong to suppose that
demonstrative colour concept possession requires such recognitional
capacities, or (as Kelly thinks) he is wrong to suppose that, as conceptualism
implies, we must have demonstrative concepts for all the shades which
feature in our experiences.
In response to this kind of concern (and unwilling to abandon his
conceptualism), Bill Brewer has proposed a rather different account of
demonstrative colour concept possession.9 According to Brewer’s proposal, a
possessor of a demonstrative concept of a shade featuring in an experience
need not have the ability to tell, after the shade no longer features in the
experience, whether or not a shade featuring in a new experience is identical
to the original shade. Rather, the concept possessor must have the ability
‘to keep track of the same shade…during a single extended period of
observation.’10
6
McDowell, 1994, p. 57.
7
Or heights, etc. Although I will focus on shades from now on, the points that I will make
apply to other properties and relations.
8
Kelly, 2001, p. 411.
9
Brewer, 2004, pp. 14-18.
10
Ibid., p. 16.
31
The key idea here then is that the demonstrative colour concept possessor
need only keep track of the relevant shade during the course of a single
experience: there is no requirement that he should be able to reidentify the
shade after a complete break in (the shade’s featuring in the) experience. And
this account, one might think, is one which is rather less likely to be shown
empirically to be incompatible with conceptualism: from the philosopher’s
armchair at least, it seems plausible that although (as a matter of empirical
fact) our capacity to discriminate colours exceeds our capacity to reidentify the
colours discriminated after a break in experience, our capacity to discriminate
colours does not (as a matter of empirical fact) exceed our capacity to keep
track of the colours discriminated over the course of a single experience.
5.
The Incompatibility of Two Claims
But if the theorist decides to accept either McDowell’s stricter constraint or
Brewer’s more liberal alternative condition on demonstrative colour concept
possession, he will need to give an account of the identity conditions for the
shades that feature in our experiences. For if a subject is to recognise that
some shade which features in a new experience is identical to a shade which
featured in a previous experience, or if he is to keep track of a particular
shade over the course of a single experience, then it must be that there is only
one shade in play: there must be only one shade that he reidentifies, or one
shade of which he keeps track.
However, I do not think that the theorist should look to the
conceptualists’ own accounts for an answer to this question, since this is a
stage at which they themselves go wrong. For in providing such accounts, the
conceptualists make two claims that are incompatible.
First, take McDowell’s suggestion that:
‘We might lay down the rule that something counts as having that shade
just in case it is indiscriminable in colour from the indicated sample.’11
Call this the identity of indiscriminables claim. So if the claim is true, in order
to possess a demonstrative concept of a shade that is featuring in my
experience:
(1)
According to McDowell, I must have the capacity to tell, after the
shade no longer features in my experience, whether or not a shade
that features in a new experience is indiscriminable from the
original shade; or,
(2)
According to Brewer, I must have the capacity to tell, at a later
stage of the same experience, whether or not a shade that is
featuring in the experience is indiscriminable from the original
shade.
But as McDowell goes on to tell us, embracing the identity of indiscriminables
claim threatens to raise a problem due to the (supposed) non-transitivity of
colour indiscriminability.12 To say that colour indiscriminability is non-transitive
11
McDowell, 1994, p. 170.
12
Ibid., p. 170-1.
32
is to say that there are shades A, B, and C where B is indiscriminable in
colour from both A and C, but C is discriminable in colour from A. Call this the
non-transitivity claim. Such a claim threatens to raise a problem if we accept
the identity of indiscriminables claim. For if both claims are true, such shades
as A and C would seem, despite their being discriminable, to have to be
identical since they are both indiscriminable from (and hence identical to) a
third shade, B.
McDowell’s proposed solution to this problem,13 which Brewer also
accepts,14 is to insist that a shade A need not be identical to a shade C just
because both shades are indiscriminable from a further shade B; rather,
shade A must itself be indiscriminable from shade C if the two shades are
identical. The reasoning here seems to be that once the theorist has ensured
that this further rule is in place, there will no longer be a danger that he will
have to count shades that are discriminable as nevertheless identical.
However, this apparent benefit comes at too great a price. For if the
theorist accepts the conceptualists’ story here, he will be forced also to accept
a wholly bizarre view of the nature of identity. For according to the
conceptualists’ proposal, we should not count shade A as identical to shade C
(since they are discriminable) despite the fact that both shades are identical to
(since they are indiscriminable from) a further shade B. But this is absurd: if
shade A is identical to shade B, then we don’t have two shades – shade A on
the one hand, and shade B on the other - rather we have a single shade
which is going by two names. And this single shade cannot be both identical
to, and different from, some further shade C.
For this reason, then, it seems to me that the apparent tension between
the non-transitivity claim and the identity of indiscriminables claim is in fact
perfectly genuine: there really is no right way to commit oneself to both.
6.
Options for the Conceptualist
This is where Graff’s arguments become relevant, since if (as I suspect) they
do indeed establish that colour indiscriminability is transitive, it is clear what
action the conceptualist should take: he should drop the non-transitivity claim.
However, if the conceptualist insists on rejecting Graff’s view and retains
his commitment to the non-transitivity claim, he will face serious
consequences: he will then need to reject the identity of indiscriminables claim
and with it both McDowell’s and Brewer’s constraints on demonstrative colour
concept possession, since if the identity of indiscriminables claim is false, both
constraints would be unsatisfiable. I shall now explain why this is.
Recall that according to McDowell’s constraint, a possessor of a
demonstrative concept of a shade featuring in an experience must have the
capacity to tell, after the shade no longer features in the experience, whether
or not a shade featuring in a new experience is identical to the original shade.
However, at best, a subject might be able to tell if a shade featuring in a new
experience is indiscriminable from a shade which featured in a previous
experience. But if the indiscriminability of two such shades does not imply
13
Ibid., p. 171.
14
Brewer, 1999, p. 175.
33
their identity, it seems impossible to see how a subject could, even in
principle, have the capacity to tell if two shades are (not merely
indiscriminable but also) identical. So if the conceptualist decides to drop the
identity of indiscriminables claim, he must also concede that McDowell’s
constraint is unsatisfiable.
Precisely the same problem applies to Brewer’s constraint: at best, the
subject of an experience in which a particular shade features might have the
capacity to tell, at a later stage of the experience, whether or not a shade
featuring in the experience is indiscriminable from the original shade. But
again, if the indiscriminability of the two shades does not imply their identity, it
seems impossible to see how the subject could, even in principle, have the
capacity at the later stage to tell if the relevant shade is (not merely
indiscriminable but also) identical to the original shade. So again, if the
conceptualist decides to drop the identity of indiscriminables claim, he must
also concede that Brewer’s constraint is unsatisfiable.
7.
Possible Objections (1)
For these reasons, then, my conclusion will be that the conceptualist should
accept Graff’s view that colour indiscriminability is transitive. In order to
establish that this conclusion is correct, however, I now want to consider two
objections to it that the conceptualist might try to make.
To anticipate, the first objector suggests that rejecting the identity of
indiscriminables claim is actually quite consistent with maintaining that
McDowell’s and Brewer’s constraints are satisfiable. The second objector, by
contrast, concedes that rejecting the identity of indiscriminables claim does
mean admitting that the two constraints are unsatisfiable, but he suggests that
demonstrative colour concept possession need not require satisfaction of
either constraint in any case.
Firstly, then, I want to investigate the prospects for the conceptualist
should he maintain that he can drop the identity of indiscriminables claim
without thereby admitting that McDowell’s and Brewer’s constraints are
unsatisfiable. One way in which he might try to do this is by conceding that
since he has rejected the identity of indiscriminables claim, he must indeed
admit that it is impossible for a subject in whose experience a particular shade
is featuring:
(1)
To tell, after the shade no longer features in his experience,
whether or not a shade that features in a new experience is (strictly
speaking) identical to the original shade; or,
(2)
To tell, at a later stage of the same experience, whether or not a
shade that is featuring in the experience is (strictly speaking)
identical to the original shade.
But the conceptualist might then claim that the (strict) non-identity of two such
shades need not stop them counting as the same for the purposes of the
proposed constraints. On this view, the constraints revolve around the idea
that in order to possess a demonstrative colour concept, a subject need not
have the capacity to tell if some later shade is strictly speaking identical to an
earlier shade; rather, he must have the capacity to tell if the later shade
34
counts as the same as the earlier shade, where the indiscriminability of the
two shades ensures that they do indeed at least count as the same.15
For after all, the conceptualist might point out, the key motivation which
underlay the imposition of the constraints in the first place was that they were
supposed to ensure that a possessor of a demonstrative colour concept must
have some bona fide ability to recognise when a shade falls under that
concept, which would in turn ensure that he can properly be said to know what
it is for something to fall under that concept. And if two shades that feature in
a subject’s experience count as the same, in virtue of their indiscriminability,
then there seems nothing to prevent that subject correctly bringing them under
a single demonstrative colour concept.
The response to this proposal is immediate, however. For if the two
shades in question are not (strictly speaking) identical, then they make a
(strictly speaking) different contribution to the content of the experiences in
which they feature. So if the two shades are brought under a single
demonstrative colour concept, this means that the features are more finegrained than the concept under which those features are brought. And this is
precisely the original difficulty which the conceptualist’s appeal to
demonstrative concepts was supposed to solve. Hence this first objection
fails: if the conceptualist decides to drop the identity of indiscriminables claim,
he must indeed accept that McDowell’s and Brewer’s constraints on
demonstrative colour concept possession are unsatisfiable.
8.
Possible Objections (2)
The second strategy that the conceptualist might pursue would involve
conceding that if he rejects the identity of indiscriminables claim he must also
admit that McDowell’s and Brewer’s constraints are unsatisfiable, but he might
then question whether it follows from this admission that possession of
demonstrative colour concepts is impossible. For there are good reasons to
think, the conceptualist might claim, that demonstrative colour concept
possession need not in fact require satisfaction of either constraint.
The idea here would be this. The original problem posed by
Wittgenstein’s case seemed to be that it placed the onus on the conceptualist
to explain why there is a disanalogy between this case, on the one hand, and
the case of the demonstrative colour concept possessor, on the other: it
seemed that the conceptualist needed to tell us what it is that the
demonstrative colour concept possessor can do, but Wittgenstein’s subject
cannot do, that ensures that the relevant concept is genuine.
But perhaps the correct response for the conceptualist to make here is
not to impose the constraints on demonstrative colour concept possession
that McDowell and Brewer do, but rather to claim that the very fact that the
15
Some of the conceptualists’ remarks suggest that they themselves might be tempted by
this view. For example, McDowell says: ‘Something that counts as having a shade
does not thereby count as a sample of the shade…’ (1994, p. 171) One way (though
not the only way) of interpreting this is as a concession that something which merely
has a shade need not be strictly speaking identical in colour to something which is a
sample of the shade. Note also Brewer’s repeated talk of ‘counting as a shade’ in
preference to talk of ‘being a shade’ (1999, p. 175).
35
demonstrative colour concept possessor’s experience features a particular
shade is itself enough to put him in a position to know what it is for something
to be that shade. (The conceptualist can happily admit that when the shade
ceases to feature in his experience, he no longer has the relevant knowledge.)
Now in Wittgenstein’s case, the subject clearly does not know what it is for
something to be ‘this tall’. Or at least, this would be so if we imagine him (say)
sitting in an armchair while performing his gestures. But in this case, the
conceptualist might point out, the relevant property does not feature in his
experience: if we rather imagine him gesturing in front of a mirror and so the
property were featuring in his experience, it is by no means clear (the
conceptualist might claim) that we would be so keen to deny that he had the
relevant demonstrative concept.
I concede that there is indeed some initial plausibility in this suggestion.
My response to it is to insist that (if the identity of indiscriminables claim is
false, as the conceptualist is currently conceding) the mere fact that a
subject’s experience features a particular shade is not enough to enable him
to know what it is for something to be that shade. For given that he cannot tell
the difference between that shade (call it ‘S1’) and some other shade, call it
‘S2’, that is indiscriminable to him from S1, he simply does not know which
shade S1 is.
Some may suspect that my response here is ineffective. For although
the subject cannot tell the difference between S1 and S2, this is not in itself
sufficient, the conceptualist might claim, to establish that he does not know
exactly which shade S1 is. For he may be able to identify S1 demonstratively,
the conceptualist might suggest, in such a way as to allow him to know exactly
which shade it is.
The conceptualist might draw an analogy with the case of identical twins:
suppose that I am looking at Amy, who has an identical twin, Becky, who
looks so similar to Amy that I would be incapable of telling them apart. Does it
follow that I do not know exactly who I am looking at? Seemingly not. If I am
looking at Amy, it would seem plausible that I do know exactly who I am
looking at, since I know that I am looking at her (as I might put it). And just as
the fact that Amy has an identical twin is irrelevant to the question of whether I
know exactly who I am looking at, the conceptualist might go on, so the fact
that S2 is indiscriminable from S1 is irrelevant to the question of whether our
envisaged subject knows exactly which shade it is that is featuring in his
experience.
I have a two-part response to this line of argument. The first part is to
suggest that the conceptualist’s current appeal to demonstrative identification
may be vulnerable to the same concerns that plagued his original appeal to
demonstrative concepts. The conceptualist’s original problem was to find a
way to assure us that his envisaged demonstrative concepts really were
genuine concepts. But his appeal to demonstrative identification at this later
stage of the dialectic threatens to generate essentially the same difficulty: we
still need a reason to believe that the supposed demonstrative identification
really is a genuine form of identification.
The second part of the response is to suggest that the conceptualist’s
analogy with the case of identical twins does not supply us with such a
reason. For although it is indeed plausible that I can identify twin Amy
36
demonstratively in a way that allows me to know exactly who I am looking at,
there is a key disanalogy between a case of this kind and the shades case.
The disanalogy consists in the fact that in the twins case, but not in the
shades case, the relevant subject satisfies some version of Brewer’s ‘tracking’
constraint on demonstrative concept possession: in the twins case, I can keep
track of twin Amy while she continues to feature in my experience; in the
shades case, by contrast, the subject cannot keep track of shade S1 while it
continues to feature in his experience.
I shall illustrate this by describing two experiments that a scientist might
perform, one involving the twins, and the other involving the shades. Suppose
that in the first experiment, I am told to keep track of Amy continuously as she
sits in front of me. At some point during the experiment, the scientist then sits
Amy’s sister Becky down beside her. The scientist then points to Becky and
asks ‘Did I tell you to keep track of her?’. My likely response to such a
question seems clear: pointing to Amy, I would surely reply ‘No – you told me
to keep track of her.’
Suppose now that the scientist tells me to keep continuous track of
shade S1, which happens to be instantiated by a particular card placed in
front of me. (We can imagine the scientist emphasising the need for me to
keep track of the shade that is instantiated by the card rather the card itself.)
Then, after a period, the scientist places a second card down next to the first,
but this second card instantiates shade S2. Pointing to the second card, the
scientist then asks ‘Did I tell you to keep track of this shade?’. Here my
answer would surely be rather different: since S2 is indiscriminable to me from
S1, I would be unable to tell that the shade instantiated by the second card is
not the shade which I was supposed to be tracking.
Now recall that according to Brewer’s constraint, in order to possess a
demonstrative concept of a shade featuring in an experience, the subject must
have the capacity to tell, at a later stage of the experience, whether or not a
shade featuring in the experience is identical to the original shade. This
condition seems not to be met in the envisaged case. For despite the fact that
the original shade, S1, has featured continuously in my experience throughout
the experiment, I am not in a position to tell that the second shade, S2, is
different from the original shade. So according to Brewer’s constraint, I lack a
demonstrative concept of the original shade, S1.
An analogue of Brewer’s constraint that applies to possession of
demonstrative concepts of objects would look like this: in order to possess a
demonstrative concept of an object featuring in an experience, the subject
must have the capacity to tell, at a later stage of the experience, whether or
not an object featuring in the experience is identical to the original object. Now
this constraint does seem to be satisfied in the twins case: so long as Amy
has featured continuously in my experience, I will immediately be in a position
to tell that any other person (even her identical twin) that features in my
experience is someone other than Amy.
Ultimately, then, the onus remains firmly on the conceptualist to explain
how it is that in the shades case, even though neither McDowell’s nor
Brewer’s constraint on demonstrative colour concept possession is met, I can
nevertheless identify shade S1 demonstratively. Certainly, the fact that
demonstrative identification is possible in the twins case, when (an analogue
37
of) Brewer’s constraint is satisfied, gives us no reason to suppose that it is
also possible in the shades case, when the constraint is not satisfied.
In the absence of any further conceptualist explanation, then, I conclude
that my original response to the current objection remains correct: if the
identity of indiscriminables claim is false, the mere fact that the supposed
demonstrative colour concept possessor’s experience features a particular
shade is not enough to put him in a position to know what it is for something to
be that shade, since it does not enable him to know which shade it is. And this
in turn casts serious doubt on the idea that the supposed demonstrative
concept is a genuine concept at all.
9.
Conclusion
I will end by recapping on the structure of my argument. I have been arguing
that the conceptualist should accept Graff’s view that colour indiscriminability
is transitive. He should do this since only by so doing can he retain a
commitment to the identity of indiscriminables claim. This in turn is essential
since only by retaining such a commitment can he accept the idea that (either
of) the constraints that McDowell and Brewer impose on demonstrative colour
concept possession are satisfiable. Finally, he does indeed need to accept
this idea since there seems good reason to think that demonstrative colour
concept possession is not possible if neither constraint is met.
References
Brewer, B. 1999. Perception and Reason. Oxford: Oxford University Press.
Brewer, B. 2004. ‘Do Sense Experiential States Have Conceptual Content?’
Forthcoming.
Campbell, J. 2002. Reference and Consciousness. Oxford: Oxford University
Press.
Evans, G. 1982. The Varieties of Reference. Oxford: Oxford University Press.
Graff, D. 2001. ‘Phenomenal Continua and the Sorites.’ Mind 110, 905-936.
Heck, R. 2000. ‘Non-Conceptual Content and the “Space of Reasons”.’
Philosophical Review, 109, 483-523.
Kelly, S. 2001. ‘Demonstrative Concepts and Experience.’ Philosophical
Review, 110, 397-420.
McDowell, J. 1990. ‘Peacocke and Evans on Demonstrative Content.’ Mind
99, 255-66.
McDowell, J. 1994. Mind and World. Cambridge, Mass.: Harvard University
Press.
Peacocke, C. 1991. ‘Demonstrative Content: A Reply to John McDowell.’ Mind
100, 123-33.
Peacocke, C. 1992. A Study of Concepts. Cambridge, Mass.: MIT Press.
Wittgenstein, L. 1958. Philosophical Investigations. Oxford: Blackwell.
38
Relational Looking and the Reversibility Constraint
Richard Price
Department of Philosophy, University of Oxford
richard.price@ccc.ox.ac.uk
In this paper I am interested in cases of relational looking; cases where one
object looks to you to bear a relation to another object. In particular, I am
interested in whether you have to see objects A and B in order for A to look to
you related to B. Here are two examples that bring out what I have in mind. In
the first example, you see two lines, A and B, before you, and you say that A
looks longer than B. In the second example, you see only one building in front
of you (that is not the Eiffel Tower), and you say that this building looks to you
the same height as the Eiffel Tower. In this paper I shall discuss the relation
between saying that A looks to you longer than B, and saying that this building
looks to you the same height as the Eiffel Tower. In particular, I will discuss
whether both statements are true in the same sense of ‘looks’.
This is an issue to do with how rich the content of experience is. When
you look at a building and say ‘this building looks to me the same height as
the Eiffel Tower’, the question is whether it is part of the content of your
experience that the building before you is the same height as the Eiffel Tower.
In this paper, I develop a constraint on relational contents of representational
states, which I call the ‘reversibility constraint’, and I argue that the case of the
building before you looking to you the same height as the Eiffel Tower does
not satisfy this constraint—hence the relation of being the same height as the
Eiffel Tower is not part of the content of experience. By contrast, when you
see lines A and B, and A looks longer to you than B, this case does satisfy the
reversibility constraint, and so it is an open possibility that it is part of the
content of experience that A is longer than B.
In the second section of the paper, I introduce a principle I call the
‘Content Principle’, which I use to define the notion of the content of
experience. I then argue that the content principle seems to entail that the
relation of being the same height as the Eiffel Tower can be in the content of
experience, even when you do not see the Eiffel Tower. I then suggest an
intuitive amendment of the content principle to resolve this puzzling conflict
between the reversibility constraint and the content principle.
In the third section of the paper, I discuss how the reversibility constraint
can do useful further work in demarcating the content of experience,
especially in contentious areas. For instance, it is a matter of controversy
whether the property of having a back, and, by extension, the property of
being a three-dimensional physical object can occur in the content of
experience. I will argue that the reversibility constraint rules out from the
content of experience the property of having a back, and, by extension, the
property of being a three-dimensional object from being in the content of
experience.
In addition the reversibility constraint rules out the property of being to
the left of me, the property of being far away from me, the property of being
circular-and-at-a-slant-from-me from the content of experience. This raises a
39
question about what kinds of location properties objects can look to have. I
discuss the options, absolutist and relationalist location properties, and I reject
both. Using the reversibility constraint, I also reject the notion that objects look
to be at certain visual angles from us. This leads to a puzzle that lies
unresolved at the end of the paper, namely what kinds of location properties
objects do look to have.
I
Consider S’s belief that A is the same height as B. This is a belief with a
relational content. When S believes that A is the same height as B, she
believes something about A, and she also believes something about B. What
she believes about B is that it is such that A is the same height as it. So, if A is
believed by S to be the same height as B, then B is believed by S to be such
that A is the same height as it. This principle holds as a matter of necessity: it
would be nonsensical to say that you believe that A is longer than B, but you
don’t believe anything about B. Necessarily, in believing that A is longer than
B, you believe that B is such that A is longer than it. This principle seems to
hold for all representational content. If A is represented as being taller than B,
then B is represented as being such that A is taller than it; i.e. one can always
reverse the order of A and B, to say how B is represented as being. Let us
express this constraint on relational contents as follows:
Reversibility Constraint:
Necessarily, if A is represented as bearing relation R to B, then B is
represented as being such that A bears R to it.
A stronger version of the Reversibility Constraint may be true as well. You
might think that, if someone believes that A is taller than B, then this is exactly
the same as believing that B is less tall than A. Or that, if one believes that A
is the same colour as B, then this is the same as believing that B is the same
colour as A. Or that, if one believes that James hit Fred, then this is the same
as believing that Fred was hit by James. The idea is that for every relation, R,
that holds between A and B, there is a reverse relation, reverse-R, that holds
between B and A. According to the stronger version of the reversibility
constraint, if you believe that A bears R to B, this is the same as believing that
B bears reverse-R to A. The weaker reversibility constraint does not require
that you believe that B bears reverse-R to A. The weaker version merely
requires that you believe that B is such that A bears R to it.
It is a difficult question whether the stronger version of the reversibility
constraint is true. There is some temptation to say that, if a subject believes
that A bears a very complex relation R to B, she may not know whether this
complex relation is symmetric or not, so she may not have a belief about
whether B bears R to A. The stronger version of the reversibility constraint
cannot accommodate this possibility. According to the stronger version, if R is
a symmetric relation, then in virtue of believing that A bears R to B, S believes
that B bears R to A. The counter-intuitiveness of this conclusion may,
however, be tempered by the fact that the stronger version of the reversibility
40
constraint does allow for the possibility that the subject does not have any
beliefs about whether R is a symmetric relation.
This puzzle is not restricted to relational beliefs. If Susanna believes that
it is not the case that Jim isn’t hot, is that the same as Susanna believing that
Jim is hot? These are difficult issues to resolve, but fortunately we do not
need to resolve them for our purposes. For our purposes, all that needs to be
true is that, if A is believed to be taller than B, then B is believed to be some
way: namely, such that A is taller than it. (I am grateful to Stephen Kearns
here. The above are developments of an objection that he put to the stronger
version of the reversibility constraint that made me think that a weaker version
was needed).
Take the example of the two seen lines, A and B, where A looks longer
to you than B. For it to be in the content of your experience that A is longer
than B, that content would have to satisfy the reversibility constraint. And that
content does satisfy the reversibility constraint. For if A looks longer to you
than B, then B looks to you to be such that A is longer than it.
But consider the case where you are looking at a building before you
and say ‘this building looks to me the same height as the Eiffel Tower’. Here
the relational looks statement does not satisfy the reversibility constraint. One
cannot say ‘the Eiffel Tower looks to me to be such that this building is the
same height as it’, because the Eiffel Tower does not look any way at all to
you at the time, since you don’t see it.
Of course, there may a sense of the word ‘looks’ in which the Eiffel
Tower looks to me such that this building is the same height as it. If this sense
exists, it is the sense in which the Eiffel Tower tends to look to me such that
this building is the same height as it. But this wasn’t the sense of ‘looks’ in the
original looks-statement. When it was said that this building looks the same as
the Eiffel Tower, this did not mean that this building tends to look to me the
same height as the Eiffel Tower—after all, this building may not tend to look to
me the same height as the Eiffel Tower.
The point illustrated in the last paragraph can be made for the beliefs
too. Imagine that the Eiffel Tower is believed by Fred to be taller than Canary
Wharf is normally believed to be. (i.e. Fred believes that: the Eiffel Tower is
taller than Canary Wharf is normally believed to be). The reversal of this
content in accordance with the reversibility constraint is not: Canary Wharf is
normally believed by Fred to be such that the Eiffel Tower is taller than it. After
all, Canary Wharf might not be normally believed by Fred to be any way at all.
Rather, the correct reversal is: the height Canary Wharf is normally believed
to be is believed by Fred to be such that the height of the Eiffel Tower is
greater than it.
One might think that when we say ‘this building looks to me the same
height as the Eiffel Tower’, what we really mean is that this building looks to
me the same height as the Eiffel Tower normally looks. The above application
of the reversibility constraint may have demonstrated that the perceptual
content does not contain the relation of being the same height as the Eiffel
Tower, but it may yet allow that the perceptual content contains the relation of
being the same height as the Eiffel Tower normally looks.
However, ‘this building looks to me the same height as the Eiffel Tower
normally looks’ does not satisfy the reversibility constraint either. The correct
reversal of the supposed content is not: the Eiffel Tower normally looks the
41
same height as this building. The correct reversal is rather: the height the
Eiffel Tower normally looks to have looks to be such that the height of this
building is the same as it. Even if it is intelligible for the heights of objects to
look certain ways to you (as opposed to the objects themselves), the problem
is the same one as before: the height that the Eiffel Tower normally looks
does not look any way to you right now, since you do not see that height.
What has been demonstrated thus far is that, when it is true that this
building looks the same height as the Eiffel Tower, and you don’t see the Eiffel
Tower, there is no experience with the content <this building is the same
height as the Eiffel Tower>. And when it is true that this building looks the
same height as the Eiffel Tower normally looks, and you don’t see the height
that the Eiffel Tower normally looks, there is no experience with the content
<this building is the same height as the Eiffel Tower normally looks>.
II
In section III, I intend to consider how the reversibility constraint can do further
work in demarcating the content of experience. In this section, however, I will
discuss a principle, that I call the ‘Content Principle’, that seems to conflict
with the conclusion of section I; that is, the content principle seems to entail
that even when you don’t see the Eiffel Tower, you can have an experience
which includes as part of its content the relation of being the same height as
the Eiffel Tower. This is puzzling, since both the reversibility constraint and the
content principle seem to be true, and yet there is evidence that they conflict.
Why do we believe that perceptual experiences have contents? What
seems to be central to the idea that experiences have contents is that there is
such a thing as an object being the way it looks, or failing to be the way it
looks. For instance a square may look triangular, and thus not be the way it
looks. The same idea makes us think that beliefs have contents: there is such
a thing as an object being the way it is believed to be, or failing to be the way
it is believed to be.
Thus, central to the idea that experiences have content is the idea that
how objects look places a condition on the objects, namely that the objects
have to be a certain way to be the way they look. Let us spell out this intuition
in the following principle:
Content Principle:
Necessarily, there is a sense of the actual sentence ‘A looks F’ which is
true iff:
(i) necessarily, if A looks F and A is not F, then A is not wholly
the way it looks.
(ii) necessarily, if A looks F and A is F, then A is partly the way it
looks.
In using the conditions in the content principle to give necessary and sufficient
conditions for perceptual content, I have assumed that the veridicality
conditions of a given experience are equivalent to the content of that
experience, and hence there are no modes of presentation in the content of
experience. I believe this assumption to be correct, though I do not have the
42
space to defend it here. In addition, the puzzle I set out below could be
rephrased in terms of the veridicality conditions of experience, rather than its
content.
There will be some expressions of the form ‘A looks F’ which do not
satisfy the two conditions in the content principle—i.e. cases where it is true to
say ‘A looks F’, even though A does not have to be F to be the way it looks.
The project of demarcating the content of visual experience is equivalent to
the project of deciding for which statements of the form ‘A looks F’ the
conditions in the content principle are satisfied. For example, the first section
of this paper has argued that, if the Eiffel Tower is not seen, the sentence ‘A
looks the same height as the Eiffel Tower’ cannot correctly report the content
of visual experience. Therefore, when the Eiffel Tower is not seen, the
statement ‘this building looks the same height as the Eiffel Tower’ will not be
true in the sense of ‘looks’ defined by the content principle: i.e. it will not be
the case that A has to be the same height as the Eiffel Tower to be the way it
looks.
In a way, this result is quite intuitive. Imagine a situation in which it is
true that A looks the same height as the Eiffel Tower, even though we don’t
see the Eiffel Tower. We wouldn’t count ourselves as misperceiving A if, in the
course of our experience, the Eiffel Tower suddenly shrunk so that it was no
longer the size of A—i.e. so that even though A looked the same height as the
Eiffel Tower, it wasn’t the same height as the Eiffel Tower. This suggests that
the truth that A looks to me the same height as the Eiffel Tower does not
require that A be the same height as the Eiffel Tower to be the way it looks. At
this stage, the results of the reversibility constraint seem to fit well with our
intuitions about what would count as a misperception of A and what would not.
However, the problem is that a respondent might object: but if the Eiffel
Tower had suddenly shrunk during the course of our experience, then from
then onwards, it would no longer be true that A looked the same height as the
Eiffel Tower. The content principle only applies to cases where A does look
the same height as the Eiffel Tower. In these cases where A does look the
same height as the Eiffel Tower, we then ask, does A have to be the same
height as the Eiffel Tower to be the way it looks? And it seems that the answer
to this is ‘yes’. For the respondent will say that if A looks the same height as
the Eiffel Tower, then if the Eiffel Tower has a height of H, it follows that A
looks H. If A looks H, then A must be H to be the way it looks; and since H is
the same height as the Eiffel Tower, it follows that A must be the same height
as the Eiffel Tower to be the way it looks. Moreover, any world in which H
equal to the height of the Eiffel Tower, but A looks H, will not be a world,
according to the respondent, in which A looks the same height as the Eiffel
Tower, and hence will be irrelevant to the question of whether the conditions
in the content principle are satisfied. The respondent will argue that the point
in this paragraph shows that condition (i) of the content principle is satisfied.
The respondent has a good case for saying that condition (ii) of the
content principle is satisfied too. If the Eiffel Tower is H, and A is the same
height as the Eiffel Tower, then that means that A is H too; given that A looks
H, it follows that A is partly the way it looks. Hence both conditions of the
content principle have been met. Since the content principle was introduced to
capture what we mean by perceptual content, it seems to entail that the
43
relation of being the same height as the Eiffel Tower is in the content of one’s
experience.
The puzzle, therefore, is that the reversibility constraint rules out the
relation of being the same height as the Eiffel Tower from being in the content
of visual experience when the Eiffel Tower is not seen. And yet the content
principle, which is used to capture the notion of the content of experience,
seems to allow that the relation of being the same height as the Eiffel Tower is
in the content of experience even when the Eiffel Tower is not seen. I shall
now suggest a possible resolution of this puzzle.
Let us say that an object, A, looks F. When considering whether A has to
be F to be the way it looks, what we must do is consider all those worlds in
which A looks the same as it does when, in the actual world, A looks F. Call
that set of worlds S1. Now we must consider all those worlds within S1 in
which A is not F, and ask whether, in all those worlds, A is or is not the way it
looks. If there are ~Fa-worlds in S1 in which A is nonetheless the way it looks,
then the conditions in the content principle have not been satisfied, and ‘A
looks F’ does not report the content of visual experience.
What is in common amongst the worlds in S1 is that A looks the same as
it does when, in the actual world, A looks F. Let us define S2 to be the set of
worlds in each of which it is true to say that A looks F. What is key to resolving
the puzzle is recognizing that S1 and S2 are not necessarily the same set.
This is because there may be worlds in S1 in which A does not look F, even
though in these worlds A looks the same as it does when, in the actual world,
A does look F. This may sound bizarre. There is an intuitive principle, that, if
two objects, A and B, look the same, then if it is true that A looks F, then it
must be true that B looks F. Call this the ‘Looks the Same Principle’.
What the puzzle brings out, I believe, is that the Looks the Same
principle is false. An object, A, might look the same from t1 to t3. At t1, it was
true to say ‘A looks the same height as the Eiffel Tower’. At t2, the Eiffel
Tower shrinks in height, so that at t3, it is not true to say ‘A looks the same
height as the Eiffel Tower’. However there is no change whatsoever in the
way A looks from t1 to t3. This example is sufficient to show that the Looks the
Same principle is false. Recognizing the falsity of this principle helps us to see
how to resolve the puzzle.
When considering whether A has to be the same height as the Eiffel
Tower to be the way it looks, when it looks the same height as the Eiffel
Tower, contrary to the content principle above, we should not consider the set
of worlds in which A looks the same height as the Eiffel Tower. Rather, we
should consider the set of worlds in which A looks the same as it does when,
in the actual world, A looks the same height as the Eiffel Tower. This set, call
it S3, will contain worlds in which A does not look the same height as the Eiffel
Tower (because the Eiffel Tower is much smaller or bigger in those worlds
compared with ours), but it looks the same as it does in worlds in which it
does look the same height as the Eiffel Tower. This is the first step to
resolving the puzzle.
In S3, there will be a world, w, in which the Eiffel Tower has a height of
½ H, whereas in our world the Eiffel Tower has a height of H. Let us say that
in w, A looks H, and is H, so it is the way it looks. w shows that the statement
‘A looks the same height as the Eiffel Tower’ does not report the content of
experience, since in w, A looks the same as it does in the actual world, A is
44
also the way it looks, and A is not the same height as the Eiffel Tower. In
general, if A looks F, and we ask whether A has to be F to be the way it looks,
we clearly have to keep the ‘look’ of A fixed. But the way to do this is not to
keep fixed all of the truth-values of looks-statements involving A. That is what
created the puzzle above. Rather, what we have to do is draw on our grasp of
what it is for A to look the same that is independent of keeping all the truthvalues of looks-statements about A fixed. This is another way of saying that
the Looks the Same principle articulated above is false. The correct
formulation of the content principle is as follows:
Content Principle*:
Necessarily, there is a sense of the actual sentence ‘A looks F’ which is
true iff:
(i) necessarily, if A looks the same as it does when, in the actual
world, it looks F and A is not F, then A is not wholly the way it
looks.
(ii) necessarily, if A looks the same as it does when, in the actual
world, it looks F and A is F, then A is partly the way it looks.
Rather than conflicting with the reversibility principle, the content principle*
supports the conclusion that, when the Eiffel Tower is not seen, one’s visual
experiences do not include in their contents the relation of A being the same
height as the Eiffel Tower. This is because of worlds such as w above, in
which A looks the same as it does when, in the actual world, it looks the same
height as the Eiffel Tower, and in which A is the way it looks, and in which A is
not the same height as the Eiffel Tower.
III
In this section I will discuss some further work that the reversibility constraint
can do in demarcating the content of experience. There is a question as to
whether experience represents things as tomatoes, tables and persons, or
whether experience represents a much sparser range of properties, for
instance colours and shapes. As the content principle illustrates, what these
questions amount to is asking whether it is ever the case that an object has to
be, say, a tomato to be the way it looks. Of course, we do say sentences such
as ‘A looks to be a tomato’, but the question is whether these sentences are
ever true in the sense that A has to be a tomato to be the way it looks.
It seems that part of what would be involved in representing an object as
a tomato would be representing the object as a three-dimensional physical
object; and part of what is involved in representing an object as a threedimensional physical object would be representing the object as having a
back.. This is where the reversibility constraint does some work. Just as the
statement ‘I have a hand’ expresses a relation between me and one of my
parts, so the statement ‘this object has a back’ expresses a relation between it
and one of its parts. For that relation to be part of the content of experience, it
would have to be reversible, i.e. if the sentence ‘this object looks to me to
have a back’ is true, then the sentence ‘a back looks to me to be such that
this object has it’ will be true. Yet, the latter sentence will not be true, since if I
45
do not see the back of the object, a back does not look any way to me at all.
Therefore the reversibility constraint entails that, when we do not see the
backs of objects, our experiences do not represent that the objects have
backs.
We often say sentences such as ‘this objects looks to the left of me’. Is it
part of the content of my visual experience that the object is to the left of me?
According to the reversibility constraint, this relation is part of the content of
experience only if one can reverse the relata in the relation and say ‘I look to
be such that this object is to the left of me’. However, if I do not see myself, I
do not look to myself to be any way at all. For similar reasons, the sentence ‘A
looks far away from me’ will not truly report the content of experience, since I
do not look to myself to be such that A is far away from me. And the same
points apply to ‘A looks in front of me’. The reversibility constraint also rules
out the sentence ‘this coin looks to me circular-at-a-slant’ from reporting the
content of experience, since that sentence is elliptical for ‘this coin looks to me
circular-at-a-slant-from-me’, which will not be reversible if I do not see myself.
One might think that the reversibility constraint allows the sentence ‘A
looks further away than B’ to report the content of experience, if A and B are
both seen. But in fact this statement is elliptical for ‘A looks further away from
me than B’, and this statement is not reversible if I do not see myself.
Similarly, even if one sees A and B, that does not mean that the sentence ‘A
looks to the left of B’ can correctly report the content of experience, since that
sentence is elliptical for ‘A looks further to my left than B’, and this statement
is not reversible if I do not see myself.
If one is unable to use properties such as being to the left of me to
characterize the locations objects look to be in, how should one characterize
these locations? One idea would be to use the notion of a visual angle. An
object might look to be at a certain angle, say 90 degrees. But being at 90
degrees is really the relation of being at 90 degrees from me (I am grateful to
Hemdat Lerman here). Hence, the reversibility constraint entails that objects
do not look to be at 90 degrees from me, if I do not see myself. This raises the
question: when we talk about objects looking to be at certain locations, what
kinds of properties do we mean?
According to relationalists, the property of being at a certain location is a
matter of standing in relations to other objects in the world. However, given
that one does not see all the objects in the world, the reversibility constraint
would rule out relationalist location properties from being in the content of
experience. According to absolutists, space is considered to be analogous to
a large container, and every point in space has an absolute xyz coordinate,
where the relevant axes are taken to be analogous to the sides of the
container. Might visual experience represent absolutist location properties? It
seems that an absolutist could not provide a full account of the location
properties in experience, for the following reason. Imagine that I am looking at
a red square on my left and a green square on my right. If I then move round
and look at them from the opposite perspective, the way things look change.
However, if experience represents absolutist location properties, then the
content of experience has not changed at all, since the objects will still look to
be in the exactly the same absolute location as they looked to be in from the
original perspective. The absolutist is unable to explain how the way things
46
look seems to change as one switches perspective on the same set of
objects.
This latter example also makes another point clear. Whatever location
properties visual experience does represent, they must be such that, when
two people are looking at the two coloured squares from opposite angles, the
location properties the squares look to have are such that the squares can
actually have them. We do not want to be forced to say that there is only one
perspective from which one can veridically perceive the squares’ locations.
The points in the above two paragraphs can be combined to specify the
two roles that must be played by the location properties that objects look to
have. Firstly, they must be able to explain how the way things look changes
as your perspective changes; secondly, the location properties that an object
is represented as having by experiences from multiple perspectives must be
capable of being had by one and the same object (i.e. experiences from
multiple perspectives must be capable of all being veridical). The puzzle is
that the properties that seem ideally suited to play these two roles are
properties such as ‘being to the right of me at such and such an angle’; and
yet, the reversibility constraint rules out these kinds of properties from being in
the content of experience. Thus, what kinds of location properties are
represented by experience? Given how closely related size and location
properties are, we can ask: what kinds of size properties are represented by
experience? These seem to me to be extremely interesting questions, and
ones that deserve some further investigation.
Before finishing I will mention two further accounts that are in conflict
with the reversibility constraint. The first is the relational view of colour
properties. Some hold that being red is the property of looking red to normal
observers in normal conditions. If one accepts the assumption there are no
modes of presentation in the content of experience (and, it seems that there is
some pressure to do this if one believes in a nonconceptual theory of
perceptual content—after all, what could these modes of presentation be
other than concepts?), then, on this view, representing something as red is a
matter of representing the thing as looking red to normal observers in normal
conditions. Yet representing this latter relation would violate the reversibility
constraint, given that you do not see normal observers whenever things look
red to you.
The second account in conflict with the reversibility constraint is John
Searle’s account of perceptual content. In order to rule out the possibility of
veridical hallucination, John Searle claimed that one’s experiential contents
have an implicit causal and self-referential component (Searle, 1983). He
claimed that, when you are looking at a yellow car before you, the content of
your experience is <there is a yellow car before me that is causing this
experience>. However, given that one does not see one’s own experiences,
the reversibility constraint would rule out from the content of experience the
relation of causing one’s experiences. One’s experiences do not look to one to
be such that the object before one causes them.
The reversibility constraint seems an essential constraint on the
relational contents of representational states. It has been shown to do much
useful work in clarifying what belongs in the content of experience and what
does not. It has also raised some very interesting questions regarding what
47
kinds of location and size properties experience represents, and these
questions will be the subject of further investigation.
Reference
J. Searle, Intentionality, 1983, CUP.
48
Perception Involving Action
Susanna Schellenberg
Pittsburg/Oxford
susannaschellenberg@hotmail.com
It is striking that only self-movers are perceivers. I will argue that it is no
coincidence. Although passivity on the part of the perceiver is an essential
feature of perceptual experience, I aim to show that perceivers are not just
passive receivers of information. The idea that perception is dependent on
action has a long history in philosophy, but has barely been elaborated in
detail. Aristotle can be read as arguing that only beings that are self-movers
can perceive. Alva Noë goes farthest with the claim that perception is
dependent on action. He writes: “What perception is … is a kind of skillful
bodily activity”.1 I do not want to go as far. However, the connection I draw
between perception and action is stronger than that perception is a means to
action and action is a means to perception. No doubt our perceptions guide
our actions and our actions facilitate us having perceptions of different
objects. Action and perception are certainly related in such an instrumental
manner. I aim to bring out, however, that the capacity to perceive is
furthermore necessarily and intrinsically dependent on the capacity to act.
Two arguments will be considered for this thesis. In Part 1, I present a
way of thinking about perception that will motivate what I will be arguing for in
the rest of the paper and thereby will lay out what I will be taking for granted.
In Part 2, I argue that the capacity to perceive objects in objective space
involves practical knowledge of how one’s perception changes as one’s
spatial relation to perceived objects changes. This could be called the
sensorimotor knowledge argument. It is subject to a host of objections, but—
as I bring out in Part 3—if one uncovers a more fundamental connection
between action and perception that this argument depends on, the objections
can be put to rest and what is attractive about the sensorimotor knowledge
argument can be retained. In Part 3, I argue that the capacity to form
sensorimotor knowledge is dependent on a perceiver being aware that she is
the acting perceiving subject. The self-awareness in play is understood
practically in the sense that a perceiver understands herself as occupying one
location in space from which she both acts and perceives. This will be called
the self-awareness argument.
Before I embark on this project, it is necessary to make two
terminological points. When I speak of action I do not have a notion in mind
that has anything to do with reason giving practices, as the notion is usually
used today in philosophical debates on action. I use the term action since that
is the term used in the lively discussion of these matters in cognitive
psychology today. I will discuss in more detail what I have in mind towards the
end of this paper. Among other things it will have to be discussed whether the
activity in question must be self-activated and whether it must be intentional. I
1
Alva Noë (forthcoming), Action in Perception (Cambridge, MA: MIT Press), p.3.
49
will take for granted that intentional actions are not necessarily actions for
reason.
As a second terminological remark, when I speak of perception, as I will
explain in a moment, I mean visual perception of objects as three-dimensional
space-occupiers. So I have a fairly rich notion of perception in mind. I hope
that my argument holds for modes of perception other than visual spatial
perception, but I will not venture into that in this paper.
I – Egocentricity Argument
Trivially, to perceive, one must perceive from a point of view. We perceive
objects from our own position in space and in relation to our own position in
space. We do not perceive objects simply to the left or to the right without
perceiving them as being to our left or to our right and in this sense we do not
just register how things are in perception, we perceive how things look from
our point of view.
If perception is relational, then a perceiver’s vantage point must play a
role in her perception. Indeed, I want to say that the perceiver’s point of view
must figure in the content of perception for the content to present itself as
perception of an objective spatial world.2 But how can the location of the
perceiver—which is simply a fact about the world—play a role at a semantic
level? In order to play a role in perceptual content, the location of the
perceiver must gain a place at a cognitive level. A first step in that direction
can be to say that in perceiving objects as spatially related to herself, a
perceiver gains awareness of her self as located in the perceived world.
Understanding egocentric spatial perception as self-locating in this way
involves an element of circularity, but a circularity that is not objectionable in
itself. At the most basic level, the ability in question involves that a perceiver
understands that she occupies the spatial location between, say, the
bookshelf and the computer. The subject of perception is thereby conceived of
as located in the world as a geometrical point of view on the world and the
location of this geometrical point is defined by reference to the egocentric
spatial content of the subject’s perceptions.
But the capacity to locate oneself is not sufficient to perceive objects as
three-dimensional space-occupiers. It is only if a perceiver can abstract from
her position in space and understand herself as one object among others that
she can gain the understanding of objective space necessary to perceive the
spatial properties of objects independently of the point of view she happens to
have in any particular situation of perception. How can this requirement be
met? In order to abstract from her point of view and gain a conception of
herself as one object among others, a perceiver must not only be able to
understand that she occupies the spatial location between, say, the bookshelf
and the computer, but furthermore needs to understand what it would mean to
occupy the spatial location that is now occupied by the computer. Another way
2
50
Christopher Peacocke (1999) can be read as developing this thought in his Being
Known (Oxford: Clarendon Press).
of expressing this thought is that perceiving objects in objective space
involves being able to imagine having alter-ego points of view and conjoining
these different possible egocentric points of view to gain an understanding of
objective space. This idea must not be understood as suggesting that there
are two kinds of space: egocentric and objective space. There is only one
space. However, the positions of objects in space can be characterized in
more or less egocentric terms. Similarly, there are not two kinds of
perceivings: perceivings from a particular point of view and perceivings of
objects as objects that fill out a certain space.
In this section, I laid out what I am taking for granted and presented a
way of thinking about perception that will motivate the argument in the rest of
this paper. In particular, I motivated the view that there is an interdependence
between how objects appear to us in egocentric space and how we perceive
them in objective space. In the next section, I will take a closer look at this
interdependence.
II – Sensorimotor Knowledge3
The fact that objects are always perceived from a particular point of view does
not challenge the objectivity of our perceptions. Our perceptions do not simply
record how things appear to us. It is more complex than that. We see the plate
as round, despite its appearing elliptical from our vantage point. How does the
way objects appear to us play a role in how our perceptions represent the
world? While the actual shape is determined by how an object fills out space, I
will understand its apparent shape as determined by its actual shape and the
perceiver’s spatial relation to the object. This is what I will call the appearance
of the object. Appearances are not mental items, but a fact about the world.
They are relational, but they are not determined by relations between objects
and their sensory effect in us, but rather by the shape of objects and the
perceiver’s location in relation to the object. Although the motivation for
speaking about appearances is to say something about perception, the
relational property that brings about an appearance exists independently of
any perceiver. An appearance of a round object just is the shape of the object
as projected on to a plane that has a specific spatial location to the round
object in question. This is very important, since it is because of this that the
view suggested will not lead into a phenomenalist view.
Although we always perceive an object from a particular vantage point,
we are able to grasp how the object looks independently of the point of view
we happen to have in any particular situation. As I argued in the previous
section, there are not two kinds of space or two kinds of perceivings.
3
The view presented in this section is a variation of a standard view in cognitive
psychology. Anyone familiar with Susan Hurley and Alva Noë’s work will realize that I
owe much of the specific formulation of the sensorimotor view to their accounts. The
view as it is described here differs from Hurley’s as well as Noë’s view in that my focus
is exclusively on relations between perception and action on a personal level. See in
particular Susan Hurley (1998), Consciousness in Action (Cambridge, MA: Harvard
University Press) and Noë (forthcoming).
51
Accordingly, the question that needs to be addressed is not how to bridge the
gap between egocentric and objective space or between how objects appear
to us and how we in fact perceive them. The question is rather how it is that
we perceive things as in objective space, although we always perceive things
from a particular point of view and how the way objects appear to us plays a
role in how our perceptions represent the world.
I aim to bring out that the capacity to perceive an object is only
intelligible together with the practical ability to spatially orient ourselves in
relation to objects. When we visually experience an object, say as round, we
do so because of implicit, practical knowledge of the way the object’s
appearance varies in the characteristic way that the appearances of round
objects vary as our relation to the perceived object changes. The details of
this point are crucial. The most modest claim is that perception involves
implicit, practical knowledge of the effects of movement on perception. A
stronger claim is to say that perceptual content is constituted by a perceiver’s
ability to exercise bodily skills. In elaborating on these ideas, it will be
necessary to gain a clearer understanding of what it means to say that an
object appears to me in a certain way. Of course the way an object appears to
us should not be understood as an epistemic intermediary between what we
encounter in experience and the perceptual content of our experience.
The idea I would like to exploit is that we perceive a plate as round
because we have encountered its elliptical appearance. It is important that this
should not be read as suggesting that we first see a plate as elliptical and only
later come to realize that it is in fact round. Nor is the suggestion that we do
not actually see plates as round. It is true that one appearance in isolation
may be misleading. But we do not learn to see round objects as round and we
do not reflect on how objects appear to us and then arrive at judgments about
how we should see them. But I do want to say that we see the plate as round
because it appears elliptical to one.4 In order to have perceptions of objects in
objective space a perceiver employs (implicit, practical) knowledge of the
effects that changes in her spatial relation to objects have on her perceptions.
More specifically, a person’s perception of an object, say a cube, is
determined by practical knowledge of the form “If I were to move to the right,
my perception of the cube would change thus and so, namely in the
characteristic way that the perception of cubes varies as a perceiver’s spatial
relation to the perceived object changes.” The grasp of such practical
conditionals between action and perception is a kind of practical knowledge.
No doubt, it would be possible to have such knowledge explicitly. But what is
involved in perception need not be explicit knowledge. I will call this practical
knowledge “sensorimotor knowledge”. It might be misleading to speak of
knowledge in this context, even if one stresses the practicality of the
4
52
The idea underlying this thought is the same idea that motivates Leibniz’s distinction
between grandes and petites representings. On Leibniz’s view our perception of a the
ocean roaring, to use his example, is constituted of a multitude of micro perceptions of
the noise that a grain of sand makes when water crashes on it. We are not aware of the
noise that every single grain of sand makes when listening to the roaring of the ocean.
Nonetheless, we hear the roaring of the ocean because we hear the noise of many
grains of sand.
knowledge. An alternative term would be “sensorimotor skill”.5 The term “skill”
suggests, however, that what is in question is gained through practice, which
is a view I would like to avoid for reasons that I will lay out shortly.
To develop the specific way that I would like to understand the
sensorimotor knowledge involved in perception, it will be helpful to think
through some central ideas of Husserl’s account of protention.6 On Husserl’s
view, perception of time is not atomized into a series of discrete instants.
Rather, our time-consciousness is a continuous flux: when listening to a tune,
at any given time, we have a ‘primal impression’ of the note that is occurring
now, note 1. When we hear the next note, note 2, we no longer have a primal
impression of note 1, but we retain it: we are aware of it as just past. As the
tune proceeds, the first note recedes further into the past and appears in ever
changing ‘retentional modifications’. Furthermore, at any given point in the
tune we ‘protain’ its future course. To say that we ‘protain’ what is to come
when listening to a tune does not mean that we hear into the future.7 Husserl
distinguishes retention from memory in that when remembering a note our
attention is directed at the note past and thus our perception of the present
phase is impaired. By contrast, when we retain a note of a tune our attention
is not directed at that note, but rather at the note that is currently to be heard.
In the very same way, Husserl distinguishes protention from expectation.
When perceiving a cube, we do not see the surface of the cube facing
away from us. As on Husserl’s account of protention, we do not direct our
attention at the surface of the cube facing away from us. Although our
attention is directed at the surface of the cube facing towards us, there is a
sense, however, in which we perceive the surface facing away from us as
well. We have expectations (expectations that do not involve actively directing
our attention at what is imagined) of what an object looks like from points of
view we do not have in the particular moment of perception. In this sense, our
perception of objects is not limited to the information projected onto the retina.
When we perceive a cube, of course, we never see the whole cube. But it is
as if we perceive the whole cube since we know that if we were to move thus
and so, we would see the cube from the other side and eventually be visually
confronted with every angle of the cube. The sensorimotor knowledge that
figures in the expectations we have of what an object looks like from other
perspectives involves entertaining the possibility of having different vantage
points to the object perceived.
This sensorimotor knowledge argument is subject to a host of objections.
I will consider three objections.
5
This is the term that Alva Noë uses to describe a similar capacity.
6
See Edmund Husserl (1905), Lectures on the Phenomenology of Internal TimeConsciousness, tr. John Barrett Brough (Dordrecht: Kluwer Academic Publishers,
1991).
7
This understanding of perception is distinct in several ways from Merleau-Ponty’s
understanding. For an engaging discussion of the latter’s views see Sean Kelly
(forthcoming), “Seeing things in Merleau-Ponty”.
53
Imaginative Mind Objection
Although one can make a case that to be a successful perceiver it helps to be
able to move around objects, why must we say that sensorimotor knowledge
necessarily figures in perception in objective space? Why not just say that we
need a concept of object and a concept of objects as solid and temporally
located three-dimensional space-occupiers in order to have perceptions of
objects in objective space. Surely, we do not need to be able to actually move
around objects in order to gain a sense of their three-dimensionality and
recognize that they appear differently from different points of view. Indeed it
would seem that all we need is an imaginative mind that can entertain the
possibility of having different points of view. I will call this the imaginative mind
objection.
The problem with this line of thought is that it requires at the same time
too much and too little. It requires too much since a perceiver does not need
the concept of an object to have perceptions of objects. No doubt an
imaginative mind can do much of the work that sensorimotor knowledge can
do, however, it seems important to take seriously that perception is a fairly
primitive cognitive skill. A perceiver needs merely the ability to perceive
objects as three-dimensional space-occupiers.
The imaginative mind objection requires too little since imagining an
object and perceiving an object with practical knowledge are two very different
activities. By replacing the practical knowledge that is in play by an
imaginative mind the presentness and particularity of perception gets lost. I
have argued that there is a sense in which I perceive the sides of an object
that are not immediately presented to me from the point of view I happen to
have in a particular situation of perception. The practical knowledge that
figures in my perception is part of my perception proper. This is a very
different idea than the idea involved in, say, having a concept of objects as
solid and three-dimensional. The difference is of the same nature as the
difference I have made with reference to Husserl between protention and
expectation. A perceiver does not direct her attention at the back side of the
object, but there is a sense in which she perceives an object as having a back
side qua perceiving it as a three-dimensional space-occupier. What I have
said so far is not sufficient to put the imaginative mind objection to rest. I will
come back to this objection in Part 3.
Sentient Statue Objection
Assuming that perception involves sensorimotor knowledge as described, is
the requirement merely that our perceptions be integrated into sensorimotor
patterns allowing us to anticipate how our perceptions would change were our
spatial relations to the perceived objects to change? If this were the case, it
would only be necessary that a person’s body can be moved in relation to
perceived objects. Or must the movement at least at times be self-movement?
The answer to these questions depends largely on how one understands the
sensorimotor knowledge in play. Noë takes a radical position, arguing that
bodily movement-perception coordination must be gained in order for
54
perceptual experience to acquire content.8 If one says that coordination of
action and perception is necessary to gain the sensorimotor knowledge
involved in perception then the ability to self-activate movement becomes
necessary for perception. Intentional movement or deliberate action play an
ineliminable role on Noë’s view, since it is only through such self-activation
that one can figure out the sensorimotor interdependence.
I do not want to go as far. Although I do not take it to be necessary to be
able to gain movement-perception coordination, I will argue in the next
section, however, that it is necessary that a perceiver have practical
awareness of being the perceiving subject, and that a perceiver is aware that
she is perceiving in virtue of controlling what she perceives through action.
From what has been said so far it is not obvious why it would not be enough
that a perceiver be moved in relation to objects. But more needs to be in play
to bring out what the ability to form sensorimotor knowledge amounts to in
order to understand what is involved in having alter-ego points of view and
conjoining these different possible egocentric points of view to thereby gain an
understanding of objects as three-dimensional space-occupiers. So in order to
understand why this sentient statue objection does not hold you will have to
bear with me some more.9
Sense Data Objection
Finally, as a last objection to the view outlined so far: why is taking the way
objects appear to a perceiver from her point of view into account not just a
way of introducing sense data? I am arguing that there is a way in which
perception presents the world as being independently of a perceiver’s vantage
point. In this respect, I perceive the plate as round. But there is also a way the
world is presented in perception that incorporates a reference to how things
appear from a perceiver’s vantage point. In this respect, the plate appears
elliptical to me. Such appearances, however, are not mental items. As I
argued above, how things appear with respect to shape is a fact about the
8
See Noë (forthcoming), p. 24.
9
A further objection against the sensorimotor knowledge view is that a necessary
condition for objective perception is for the perceiver to be in a position to regard
diverse perceptions as perceptions of a single enduring and distinct object. In order to
have perceptions of a single enduring and distinct object a perceiver must be able to
recognize two distinct perceptions as successive perceptions of the same object and
distinguish this case from cases in which successive perceptions are of two different
objects. As was argued above, sensorimotor knowledge allows us to recognize any
particular appearance of an object as only one of many possible ways that an object
can present itself to a perceiver. But this thought is not the same as the thought that
diverse perceptions are recognized as perceptions of a single enduring object.
Sensorimotor knowledge as the idea has been unraveled so far cannot account for
the capacity to distinguish between successively perceiving one and the same object
and successively perceiving qualitatively indistinguishable but numerically distinct
objects, because in order to regard successive perceptions as perceptions of the same
object, a perceiver must be able to ascribe them to a numerically identical subject
whose route through the world anchors them to a single object. For a helpful discussion
of related ideas see Quassim Cassam (1997), Self and World (Oxford: Oxford
University Press).
55
object and the vantage point that the perceiver happens to have on the
perceived object. Therefore, the appearances that I am arguing are crucial for
an understanding of visual spatial perception have nothing to do with sense
data. The thought that I think must be taken very seriously in an account of
perception is that perception does not just represent the objective spatial
properties of objects; it represents how things are in relation to the perceiver.
But taking this thought seriously must not be understood as introducing
sense data. Accepting that there is a mutual dependency between apparent
shapes and actual shapes need not involve understanding perception as a
process of constructing an internal representation. The appearance of an
object from a perceiver’s point of view is difficult to isolate from how we
perceive the object on the basis of our sensorimotor knowledge. It is only
when we, for instance, learn to draw realistically that we train our eyes to see
that a plate appears elliptical from most points of view.
When we take a realistic painter’s point of view, we perceive the round
object as elliptical. We perceive the object as elliptical because we are
confronted with a round object, rather than perceiving a round object because
we encounter an elliptical appearance. In both cases, there is an
interdependence between apparent shape and actual shape. Whereas in the
first case one abstracts from one’s vantage point and perceives the shape of
the object independently of one’s point of view, in the second case one brings
one’s vantage point into the picture and perceives the shape of the object as it
appears to one from one’s point of view.
To say that there is an interdependence between apparent shape and
actual shape does not mean that one is necessarily consciously aware of both
the apparent shape and the actual shape. And to say that the actual shape
and the apparent shape both play a role in perceiving the shape of an object
does not involve saying that there is an inconsistency in the perceptual
content, since we do not perceive the round object both as elliptical and as
round, but rather as either appearing elliptical from our point of view or as
being round.10
III – Self-Awareness Argument
I have argued that perception involves sensorimotor knowledge. I aim to show
now that the capacity to form sensorimotor knowledge is dependent on a
more fundamental connection between action and perception, namely that a
10
56
I take the interdependency between apparent shape and actual shape to be an
instance of the interdependence between how an object appears to one in perception
and how the object actually is; an interdependence that is involved in modes of
perception other than visual spatial perception. A difficult case is the perception of
color, since the color of a surface is not independent of its appearance in the way the
actual shape of an object is. Unlike shapes, one might argue, colors are themselves
appearances. When perceiving a wall it might be just as correct to say that the wall
appears uniformly white, as it is correct to say that the wall appears to have patches of
pink and green. But here again there is an interdependency, in this case, between two
ways that the wall appears to one:
perceiver be aware that she is the acting perceiving subject. Only through
such awareness can we make full sense of what it means to understand
ourselves as one object among other objects (egocentricity argument) and as
keeping track of our perceptions as our spatial relations to perceived objects
change (sensorimotor knowledge argument). Uncovering this dependency on
self-awareness will put the objections that can be raised against the two
previous arguments to rest and allow us to keep hold of what is attractive
about the two arguments. Furthermore, it will bring out how the egocentricity
argument and the sensorimotor knowledge argument are connected. Keeping
track of how our perception changes as our spatial relation to perceived
objects changes requires awareness of one’s own position in space in so far
as this position is the point of origin of our actions and perceptions.
If we take seriously the idea that how things look from here is a relational
property that figures in the content of perception, then the perceiver’s vantage
point must figure in the content of perception. Encountering the elliptical
appearance of the plate is what allows us to perceive the plate as round, but
only because the vantage point from which we perceive the plate enters in our
perceptual content. So the subject of perception plays a role in the content of
her perceptions to the extent that she forms the point of origin of an
egocentric frame of reference. Furthermore, to play a role in perceptual
content, the perceiver’s spatial location must gain a place on a cognitive level.
(These are two of the ideas that I laid out in the first part and am taking for
granted in the context of this paper.)
The location of the perceiving subject as the egocentric frame of
reference, I will argue, can figure in our perception only in so far as we are
aware of ourselves as acting beings. The idea I am trying to exploit is that
perceivers have an understanding of their location in space, because it is the
location from which they both perceive and act. Perception is essentially
perspectival insofar as perceptual content is structured in subject-dependent
terms. But the possibility for action that is involved in the egocentrical
organization of perceptual content allows us to go beyond the perspectival
representation of objects and to perceive them in objective space. So
paradoxically, it is the egocentricity of perception that allows us to transcend
our perspectival frame of reference.
It is important that both the capacity to perceive and the capacity to act
in play are understood on a personal level. In this respect, the view I am
suggesting differs from an argument that takes subpersonal interrelations
between perception and action into account. Susan Hurley takes such an
approach in her Consciousness in Action, arguing that the interdependence
between the contents of intentions and of perception on a personal level can
be understood as emerging from the codependence of perception and action
on dynamically circular subpersonal relations. In particular, she argues that
feedback from motor outputs to sensory inputs plays a critical role within such
a subpersonal dynamic system. The point I am making is fundamentally
different in that the focus of the interdependency between action and
perception is on the perceiving subject, not its subpersonal system. But saying
that there is an interdependence on a personal level need not involve denying
that there is an interdependence on a subpersonal level; I am simply not
taking a stance on it.
57
It must be added that Hurley and similarly Noë in his forthcoming book
Action in Perception self-consciously slip back and forth between personal
and subpersonal levels. Noë argues that bodily activity and the physical
implementation of perception in the brain and nervous system are
epistemologically on the same level of investigation and, thus, rejects any
autonomy thesis, any claim that a philosophical analysis of perception is
epistemologically independent of a scientific analysis. I believe that one must
hold on to a distinction between bodily movement and the processes of an
organism’s nervous system that is not just a distinction in observability of the
movement involved. But this is a big topic. In this paper, I will take for granted
that there is a clean distinction between perception on a personal level and
the processing at the concrete implementation of perception.
Now if the capacities to perceive and act are understood on a personal
level, why does this account of self-awareness involved in perception not
introduce a superfluous and potentially problematic intermediary stage in
perception? This is a version of the sense data objection that I argued against
in the previous section. It is important to keep in mind that when perceiving
external objects the self is not the object of attention. Perceptual attention is
focused on the objects perceived, not on the perceiver or the mental state of
perception. So the self-awareness in play must not be understood in analogy
to the awareness involved in perception of objects.
In this sense, the view presented here differs fundamentally from what
could be called the experiential knowledge view. Awareness of what it feels
like to be perceiving or any other form of introspection or awareness of one’s
inside is not what constitutes the essential awareness of oneself as the point
of origin of perception. I neither feel that I am perceiving, nor is there
something that is perceiving that I then realize is me. One’s perceptions
cannot be experienced as unowned or of uncertain ownership.11 Indeed, the
question whether awareness and felt ownership of perception are separate
issues does not make much sense. Although one can imagine cases in which
the thought that “I found myself perceiving x” makes sense, the situation in
which one is aware that one is perceiving but unwilling to ascribe one’s
perception to oneself does not make sense. By understanding the selfawareness in play as immediate and non-relational in this way, I am rejecting
the view that awareness of an object and awareness of myself as perceiving
are two states of mind that in perception successfully come together.
How do these ideas of the ownership of perception relate to the thesis
that the capacity to perceive is dependent on the capacity to act? I am arguing
that a perceiver is aware that she is perceiving because she is aware of
occupying one space from which she both perceives and acts. But why would
perception alone or action alone not be enough to bring about the awareness
of one’s own location in space that is said to be necessary for perception of
objects in objective space? Why does one need the two?
11
58
There are interesting analogies between the sense of ownership involved in perception
and the sense of ownership involved in action. For a discussion of cases in which
sense of action and sense ownership can come apart see Anthony Marcel and
Christopher Peacocke’s (2003) discussion of the Anarchic Hand syndrome in Naomi
Eilan, Johannes Roessler (eds.), Agency and Self-awareness (Oxford: Oxford
University Press), pp. 48-110.
The answer to this question leads to the idea that spatial concepts are
not simply related to the place we occupy, but rather to the specific way we
occupy that space concerning our possibilities for action. We need at least an
understanding of what it would mean to reach out to the glass to perceive the
glass as within reach and we need to have an understanding of what it would
mean to move our body upwards in order to have an understanding of the
spatial concept of up.12
The idea behind the thought I am interested in is that a perceiver is
aware that she is perceiving in virtue of controlling what she perceives through
action.13 When we perceive a cube we know that if we were to move thus and
so, we would see the same cube from another angle. We are aware that we
are perceiving rather than imagining an object because when we, say close
our eyes, we cease to see the object and when we make a step to the left our
perception changes in expected ways. Action alone or perception alone is not
sufficient to gain the self-awareness that is said to be necessary for
perception.14
Overintellectualization Objection
Now why does understanding the awareness that is said to be involved in
perception along these lines not run the risk of over-intellectualizing
perception? In reply to this question it needs to be insisted that the notion of
self-awareness in play is understood in a radically practical way. The point of
origin figuring in perception need not be conceptualized in any conceivable
manner and the self-awareness in play need not involve information about the
self, its states or their contents. A perceiver need not have the capacity to
have first-person thoughts, nor does she need the capacity to ascribe
12
The idea I am interested in is related to Evans’s thought that “an egocentric space can
exist only for an animal in which a complex network of connections exists between
perceptual input and behavioural output” (See his (1982), Varieties of Reference
(Oxford: Oxford University Press), p. 154). Spatial concepts are not simply related to
the place we occupy, but rather to the specific way we occupy that space concerning
our possibilities for action. If I am tilting my head, I do not see objects on the verge of
sliding off the surface of the earth. The reference of ‘up’ is not determined by the
direction of my head, but rather by how I would move my body, given what I perceive. It
is because our perspectival perceptions involve how we would move and act that
perceptual content gains objectivity through its egocentrical structure. In this sense, the
self enters the content of perception as the point of origin of an egocentric frame of
reference only in so far as we understand ourselves as acting beings.
13
This way of thinking about the self-awareness in play is similar to the basic insight of
the ecological understanding of agency according to which perceptual information
involves self-awareness in virtue of its role in controlling actions. For a critical
discussion of the ecological sense of self-awareness along Gibsonian lines, see
Johannes Roessler (2003), “Intentional Action and Self-Awareness” in Eilan, Roessler,
pp. 383-406.
14
This thought is related to a reading of G.E.M Anscombe (1957) according to which
there are important connections between action control and knowledge of actions “from
within”. Anscombe can be read as arguing that an agent is aware of what she is doing
in virtue of controlling her action, rather than on the basis of observation or
introspection. The notion of control, involved in perception is naturally of a very different
kind. See her Intention (Cambridge MA: Harvard University Press).
59
perceptions to herself. Furthermore, the suggested view is not that a perceiver
locates her own position in space by reflecting on her spatial relation to
perceived objects. As I argued above the self-awareness in play is understood
as immediate and non-relational. A perceiver is aware of her location in a dual
mode: the zero-point of perception presents itself as being the zero-point for
bodily movement. One occupies one position from which one both perceives
and acts.
When a cat perceives a chair, it might not see it as a chair, but it sees
something that is located in a certain relation to itself and something that it
can either choose or choose not to jump onto. Through perception it gauges
the distance it must jump to land on the chair. Its location in relation to the
chair must figure in its perception for it to be able to flex its muscles such that
it ends up landing on the chair.15
Imaginative Mind Objection and Sentential Statue Objection revisited
Against the sensorimotor knowledge view the objection was raised that a
perceiver must only be able to imagine having different points of view in order
to acquire the sensorimotor knowledge necessary for objective perception. If it
is not necessary that a perceiver actually has different points of view, it is
unclear why a perceiver need to move in space or think of herself as being
capable of moving in space to acquire sensorimotor knowledge. I called this
the imaginative mind objection.
An objection could be raised on similar grounds against the selfawareness argument. I have argued that a perceiver needs the capacity for
self-movement, but it is not clear why a perceiver must ever actualize this
capacity. Furthermore, why not say we need perception, sensation, and
cognitive abilities that allow us to have self-awareness and imagine occupying
different locations in space. Why do we need the capacity to act at all? Finally,
surely the connections between perception and action are not so tight as to
exclude the possibility that a being that is not capable of self-movement can
perceive objects as in objective space. So, why does bringing self-awareness
into play refute the imaginative mind objection?
In order to discuss this question, it will be necessary to take a closer look
at the notion of action in play. The notion of action in play in the selfawareness argument differs from the notion of action in play in the
sensorimotor knowledge argument. The question how to understand the
action on which perception is claimed to be dependent must therefore be
answered differently with respect to the different ways in which perception is
dependent on bodily activity.
As I argued in the last section, it is not obvious that actual token actions
need to be involved for perceivers to have sensorimotor knowledge. What
constitutes sensorimotor knowledge are not token actions, but rather practical
knowledge of how perceptions would change, were the perceiver to change
15
60
The example aims only to exemplify that the self-awareness in play is understood
practically in a way that is unproblematic to ascribe to cats. The example is not meant
to make any implications about cat-perception. It cannot be taken for granted that cats
have perceptions of objects as three-dimensional space-occupiers.
her visual angle on the perceived object. By contrast, the self-awareness
argument was formulated in terms of capacities to act. But again, it is not clear
that actual actions need to be involved. Certainly a subject who is temporarily
unable to act can have a full-fledged conception of her location in space from
which she perceives and from which she would act were she able to act.
On the view suggested a perceiver must, however, know what it means
to, say reach out to an object, when perceiving the object as within reach. It is
not required that a person actually reaches out or has reached out in the past.
But what does it mean to say that a perceiver knows what it means to act,
what it means to reach out to something? Certainly it does not mean knowing
what it feels like to reach out to something or any other form of proprioception.
What is meant is rather knowing what it takes to reach out to something or
knowing what the success conditions are to reach out to something. Since we
must know what it would mean to act on our perceptions, being just moved in
relation to objects, as say plankton are, is not sufficient to have perception of
objects in objective space. This is why the sentient statue objection does not
hold. But the alternative is not that the movement be self-activated. Rather I
am taking a different line altogether, one that involves minimal cognitive skills,
namely knowing what it would mean to act.
But why does this not mean that it would be enough to say that a
perceiver need to be able to imagine herself to be able to act? The answer to
this question depends on what one means by imagining oneself to be able to
act. If knowing what it takes to do this or that counts as imagining oneself as
being able to act, then indeed that is all that is required. But this is not how the
imaginative mind objection was formulated.
The objection was that having a concept of object and a concept of
objects as solid and temporally located three-dimensional space-occupiers is
sufficient to have perceptions of objects in objective space. Having such
concepts is not sufficient to know what it would mean to do this or that, say,
reach out to something that one perceives as within reach. In this sense, as I
argued in the last section, the imaginative mind objection requires at the same
time too much and too little.
IV – Conclusion
In Part 1, I laid out what I am taking for granted, namely that perception is
relational, that there is such a thing as perception of objects as threedimensional space-occupiers, and that in order to perceive objects as threedimensional space-occupiers a perceiver must have an understanding of
objective space. To have such an understanding a perceiver must be able to
abstract from her own point of view and understand herself as one objects
among others.
The fact that objects are always perceived from a particular point of view
does not challenge the objectivity of our perceptions. We are able to perceive
the shape of objects independently of the point of view we happen to have in
any particular situation. In Part 2, I argued that our perception of an object is
determined by practical knowledge of how our perception changes as our
spatial relation to the perceived object changes. This sensorimotor knowledge
61
brings out the interdependence between the apparent shape and the actual
shape of the object. While the actual shape of an object is determined by how
the object fills out space, its apparent shape is determined by the actual
shape of the object and the perceiver’s spatial relation to the object. The
apparent shape, what I have called the appearance of the object, is not a
mental item, but rather a fact about the world.
The egocentricity argument as well as the sensorimotor knowledge
argument is subject to a host of objections. In Part 3, I argued that if one
uncovers a more fundamental connection between action and perception that
these two arguments depend on, the objections can be put to rest and what is
attractive about both arguments can be retained.
Both arguments require that a perceiver be aware that she is the acting
perceiving subject. Keeping track of how our perception changes as our
spatial relation to perceived objects changes requires awareness of one’s own
position in space in so far as this position is the point of origin of our actions
and perceptions. If we take seriously the idea that how things look from here
is a relational property, then we must take seriously the idea that the subject
of perception plays a role in the content of her perceptions to the extent that
she forms the point of origin of an egocentric frame of reference: perceptual
content is organized egocentrically representing the perceived object as being
in a certain spatial relation to the perceiver. Encountering the elliptical
appearance of the plate is what allows us to perceive the plate as round, but
only because the vantage point from which we perceive the plate enters into
the content of our perception.
When I say that the self enters the content of perception as the point of
origin of an egocentric frame of reference, I do not mean that a perceiver
being aware of herself as an acting perceiver is what figures in the content of
perception. It is rather the spatial location of the perceiver in relation to the
perceived object that figures in perceptual content. But this spatial location
can only figure in perceptual content because the perceiver is aware of herself
as the acting perceiving subject. Perceivers understand themselves as
occupying one space from which they both perceive and act and this selfunderstanding is crucial for perception in objective space, because it brings
about the notion of self that figures in the content of perception and allows us
to transcend the egocentricity of our perception.
62
The Rationality of Deductive Inference
Declan Smithies
NYU/Warwick
dgs224@nyu.edu
1.
The Question
Deductive inference is one of the ways in which it can be rational to form
beliefs. What makes it the case that it is rational to form beliefs in this way?
And in particular, what is the role of consciousness in explaining why it can be
rational to form beliefs in this way?
2.
Fodor’s Language of Thought Model
I want to approach this question by considering the Fodorian idea that the
language of thought hypothesis explains the rationality of deductive inference.
More precisely, I want to consider the claim that the language of thought
hypothesis provides a constitutive explanation of the rationality of deductive
inference, rather than merely a causal explanation in terms of “enabling
conditions”.1
The explanation takes the following shape: (1) beliefs (and other
propositional attitudes) are identified with physical structures that are causally
efficacious; (2) deductive inference consists in the occurrence of causal
transitions between these physical structures; (3) the causal transitions
between them are sensitive only to their formal (i.e. non-semantic) properties;
(4) there is an isomorphism between the formal properties of these physical
structures and the semantic properties of their propositional contents; (5) the
causal transitions between these physical structures are sensitive to their
formal properties in such a way as to preserve relations of truth-functional
consequence between their propositional contents.2 Fodor summarizes this
line of explanation in the following passage:
We can therefore build machines, that have…the following property:
The operations of the machine consist entirely of transformations of
symbols;
In the course of performing these operations, the machine is
sensitive solely to the syntactic properties of the symbols;
1
In fact, Fodor himself maintains what he calls a “meretricious metaphysical neutrality” in
his writings (e.g. 1998, Ch.1); but with respect to adopting the stronger, constitutive
commitment, he is prepared to say, “...you could do so if you were so inclined” (1987,
p.156 fn.6).
2
Fodor (1987) does not make it clear why an explanation of the rationality of deductive
inference requires the language of thought (LOT) hypothesis, rather than merely what
Fodor calls the representational theory of mind (RTM). (The issue is whether the formal
properties of the relevant physical structures are required to have constituent syntax.)
However, in Fodor and Pylyshyn (1988), there is an argument that LOT is required to
explain the systematicity of deductive inference.
63
And the operations that the machine performs on the symbols are
entirely confined to altering their shapes
Yet the machine is so devised that it will transform one symbol into
another if and only if the propositions expressed by the symbols that are
so transformed stand in certain semantic relations – e.g. the relation that
the premises bear to the conclusion in a valid argument. (Fodor 1987,
p.19)
Intuitively, however, we could build one of Fodor’s machines without thereby
building a rational thinker capable of engaging in genuine deductive inference.
If this intuition is right, then the proposed conditions are not sufficient to
explain the rationality of deductive inference. What, then, does this account
leave out?3
3.
Searle’s Chinese Room
This question is closely related to the one that Searle considers in his famous
article, “Minds, Brains and Programs”. Searle’s question is whether purely
computational conditions (of the kind that Fodor articulates) are sufficient for
genuine intelligence, which we can gloss in terms of rational thought. The
conclusion that he reaches on the basis of the Chinese Room thoughtexperiment is that purely computational conditions are not sufficient: the man
in the Chinese Room satisfies all the relevant computational conditions, but he
fails to understand the meanings of the Chinese symbols he is processing,
and so the processing is not genuinely intelligent.
Searle’s diagnosis is that what is lacking in this case of the Chinese
Room, as well as in many other cases of so-called “artificial intelligence”, is
the presence of genuine intentional content:
Because programs are defined purely formally or syntactically, and
because minds have an intrinsic mental content, it follows immediately
that the program by itself cannot constitute the mind.
In later work (Searle 1990, 1992), he goes on to fill out this conception by
arguing (1) that genuine, as opposed to merely “as if”, intentionality requires
consciousness, in the sense that every genuinely intentional state must be
available to consciousness; and (2) that the satisfaction of purely
computational conditions does not suffice for availability to consciousness. It
follows from these claims that the satisfaction of purely computational
conditions does not suffice for genuine intentionality. And given that genuine
intentionality is required for genuine intelligence, it follows that the satisfaction
of purely computational conditions does not suffice for genuine intelligence.
A number of comments are relevant here. First, it is not at all clear that
all the relevant computational conditions have in fact been met in Searle’s
case. For instance, he does not consider the point that genuine intelligence
must be structured and systematic in the right kind of way.4 As he describes
3
It is worth noting that Fodor’s proposed explanation is reliabilist to the extent that he
takes it to be sufficient for an explanation of the rationality of deductive inference that
one should explain its distinctive kind of reliability, i.e. its validity.
4
Compare Block (1980).
64
the case, the competence of the Chinese Room is due to its occupant’s use of
a mere look-up table. Suppose, then, that we adjust the case so that various
states of the system make a systematic causal contribution to the behavioural
responses exhibited by the Chinese Room. Now we can challenge Searle’s
contention that genuinely intentional states must be available to phenomenal
consciousness. For instance, we can argue that intentional contents need to
be ascribed to the states of the system in order to explain the relational
properties of its behavioural responses, regardless of whether or not those
states are available to consciousness.5 On this line, the ascription of
intentional content to states which are not available to phenomenal
consciousness need not be a purely instrumentalist matter, but may be
answerable to perfectly objective constraints on explanation.
It seems to me that Searle has misdiagnosed the intuitive significance of
his own case. What is intuitively lacking in the Chinese Room is not
intentionality per se, but more specifically understanding of the Chinese
symbols being processed. The claim that understanding requires
consciousness (whether we consider linguistic understanding or merely
nonlinguistic conceptual understanding) is much more plausible than Searle’s
own claim that intentionality requires consciousness. After all, there is a
compelling intuition that a mere machine without any conscious states could
not be credited with any kind of linguistic or conceptual understanding, even if
it could be attributed intentional states of some other kind. But some kind of
linguistic or conceptual understanding is surely a prerequisite for forming a
belief rationally on the basis of deductive inference. Since the Fodorian
account of the rationality of deductive inference is silent on the issue of
consciousness, this suggests the beginning of an answer to the question of
what that account leaves out. However, it is no more than a beginning: the
claim that understanding requires consciousness needs both a sharper
formulation and a principled explanation.
4.
Brewer’s Non-Reflective Access Requirement
In the following passage, Brewer makes an intriguing suggestion about why
consciousness might be a requirement for rational deductive inference:
Any purely mechanical dispositional account of the matter is
unacceptable. To start to see why, consider what is involved in following
a valid deductive argument with real understanding, in a way that yields
knowledge of its conclusion…. In cases where I do not go wrong, I am
correctly compelled by the argument to believe its conclusion. In
following and fully understanding the argument, this compulsion is not
simply a blind and mysterious manipulation of my beliefs by some
reliable mechanism, however well established by evolution, benevolent
hypnosis, or whatever. I am not just a machine which runs along those
rails. For if my following the argument is really to extend my knowledge,
then my understanding of it must give me some appreciation of why I am
right in believing its conclusion. I have to have some grip on how I
5
Compare Peacocke (1994).
65
thereby know the conclusion, and my belief should be guided by this
understanding. A disposition to take beliefs on board in parallel with the
steps of the argument is on its own insufficient for the argument to
provide me with genuine knowledge. For such beliefs would come as a
succession of mere hunches, wholly unsubstantiated for me by the de
facto validity of the argument propelling my endorsement of them.
(Brewer 1995, p.242)
The suggestion is that forming a belief on the basis of deductive inference in a
way that is rational, and thus a potential means for extending one’s
knowledge, requires understanding the argument that one is following.
Understanding the argument, in turn, involves an appreciation of how the
conclusion follows from the premises, and hence an appreciation of why it
would be rational to believe the conclusion, given belief in the premises. And
since such an appreciation requires consciousness, it follows that genuinely
rational deductive inference requires consciousness.
More specifically, Brewer proposes that for a belief to be rationally
formed on the basis of deductive inference, the formation of the belief must
actually be guided by a conscious appreciation of why it would be rational to
form it, which is provided by an understanding of the argument in question. If
the belief is not formed on the basis of any such appreciation, then the
suggestion is that it will seem, from the subject’s own perspective, to be no
different from blind guesswork or an ungrounded hunch. This, in turn, prompts
an intuition that it can be nothing more than a matter of mere luck, from the
subject’s perspective, if the belief is true. But then, of course, the belief cannot
be rational. This point emerges in the following passage:
More generally, the problem for any purely mechanical dispositional
account is that it is bound to ignore this sense of why one is right in
exercising the capacity in question as one does, which is crucial if this is
to make cognitive contact with the truth in the relevant area. It is the fact
that the capacity has as its point ascertaining the truth on some
matter…which sets up the norms for its correct exercise. Some
appreciation of how what one is up to is onto this truth, is sensitive to the
resultant norms, is therefore essential if exercising it is to be more than a
blind mirroring of the norms, extrinsically, and in this sense only
incidentally, in contact with the truth. (1995, p.243)
So, in general, if a belief is to be rational, then it must be formed in a way that
is guided by some conscious appreciation of why it would be rational to form
it.
It is on the basis of such considerations that, in later work, Brewer
imposes the following generalized access requirement on rational belief and
action:
If a person’s reasons are to be cited as her reasons for believing or
doing what she does, then she necessarily recognizes them as such.
(1999, p.166)
And moreover:
…she is guided in making the transition by her recognition of her reason
as a reason for doing so. (1999, p.165)
66
The difficulty here is to see how to avoid a familiar kind of regress problem. In
fact, there are two kinds of regress problem to be avoided. Suppose that the
way in which the subject recognizes her reason-giving states as such is by
forming second-order beliefs about them to the effect that they give her
reasons for belief or action. Now, these second-order beliefs must themselves
be formed in a way that is rational. But then the reason-giving states on which
these second-order beliefs are based must themselves be recognized as
such, which requires the subject to form third-order beliefs about these states
to the effect that they give her reasons. And these third-order beliefs, too,
must be rational. So, we are embarked on an infinite regress. Perhaps the
regress is not strictly vicious, but it does have the vastly implausible
consequence that the rationality of a single belief depends on an infinite
hierarchy of higher-order rational beliefs.
However, we get a regress which is more clearly vicious if the only
plausible account of the rationality of higher-order beliefs makes it dependent
on the rationality of beliefs lower down the hierarchy. But certainly in the
logical case, this can seem to be the only plausible account. For we might
suppose that recognizing the rationality of a belief based on modus ponens
requires recognizing that modus ponens is a valid rule of inference. And we
are supposing that this recognition must be a matter of rationally formed
belief. But the rationality of a belief that modus ponens is valid must surely
depend on making a modus ponens step, as in the following piece of
reasoning:
An argument is valid just in case, necessarily, if its premises are true,
then its conclusion is true. Now, suppose that p is true and that if p, then
q is true. Well then q must be true, as a matter of necessity, whatever p
and q stand for. So, an argument of the form – p; if p, then q; therefore, q
– is valid.
However, the rationality of making such a modus ponens step is precisely
what we are trying to explain; so it can seem that any purported explanation
along these lines is guaranteed to presuppose what it is trying to explain.
Brewer’s strategy for avoiding such regress problems is to propose a
distinctive account of the way in which the access requirement is satisfied.
According to his proposal, the access requirement need not be satisfied at the
reflective level by means of higher-order beliefs about one’s reason-giving
states; rather, it can be satisfied at the unreflective level by means of one’s
reason-giving states themselves. On this view, it is an essential feature of
reason-giving states that being in them suffices for satisfaction of the access
requirement; no further, reflective thoughts about those reason-giving states
are required. So, his version of the access requirement is a non-reflective
access requirement.
Unfortunately, it is not clear how to make sense of the claim that reasongiving states embody a built-in conscious recognition of their own reasongiving status. This seems to require a kind of self-referentiality in the contents
of reason-giving states which is obscure, at best, and possibly even
incoherent.
In fairness, Brewer’s (1999, Ch.6) account of the rationality of
perceptual-demonstrative judgements suggests that he has something less
demanding in mind. According to this account, perceptual experiences provide
67
reasons for perceptual judgements about the external world because they
embody an appreciation on the part of the subject that the experience is
causally dependent on the state of the external world in relevant respects. It is
in virtue of embodying such an appreciation that perceptual experiences
provide reasons for perceptual judgements. It could be objected that this does
not suffice, strictly speaking, for the claim that perceptual experiences embody
an appreciation of the fact that they themselves provide reasons for
perceptual judgements. But it could equally be replied that, loosely speaking,
it suffices for the claim that perceptual experiences embody a sense of “why it
would be right” to make perceptual judgements on their basis.
The analogous claim in the case of deductive inference would be that a
thinker’s understanding of the premises and the conclusion of an argument
embody a conscious appreciation of the fact that the conclusion follows from
the premises. Loosely speaking, this can be described as a sense of “why it
would be right” to believe the conclusion of the argument on the basis of
believing its premises, even in the absence of an explicit recognition of the
fact that my belief in the premises provides me with a reason to believe the
conclusion. The proposal, then, would be that the rationality of deductive
inference depends on the fact that it is made on the basis of a conscious
appreciation of the fact that the conclusion of the argument follows from its
premises. This proposal has recently been endorsed by Christopher
Peacocke, who expresses it as follows:
How can we elucidate the rationality of the thinker’s judgements? One
intuitive account is that in making a rational transition to a judgement that
p [on the basis of deductive inference] a thinker must know what it is for
it to be true that p, must appreciate that his grounds or reasons for the
transition to the conclusion that p suffice for the truth of p, and must be
making the judgement because of his appreciation that these grounds or
reasons so suffice. (2003, p.176)
However, neither Peacocke nor Brewer says much about how the relevant
kind of conscious appreciation might be constituted so as to play this role in
grounding the rationality of deductive inference. But this issue turns out to be
crucial in the evaluation of the proposal.
5.
Boghossian’s Challenge
Suppose that appreciating the validity of a form of argument is a matter of
explicitly articulating the form of the argument and rationally judging it to be
valid. If this is what it is involved, then there seem to be compelling objections
against the claim that a deductive inference is rational only if the subject
makes the inference on the basis of some such appreciation of its logical
validity. The first objection is that the requisite conceptual capacities are
lacked by most normal adults, not to mention animals and children; but it
would be intuitively implausible to deny on this basis that they have any
capacity for rational deductive inference. The second objection we have
already considered, in effect. If we suppose that the rationality of judgements
about the validity of forms of argument itself depends on rational deductive
inference, then it follows that the rationality of deductive inference cannot
68
depend on an appreciation of the validity of a form of argument; rather, the
direction of dependence must be the reverse.6
One line of response to these objections would be to claim that we have
some kind of direct and non-inferential, quasi-perceptual access to the validity
of forms of argument. The claim would be that such a quasi-perceptual faculty
of “rational intuition” is possible in the absence of a sophisticated capacity for
articulating general forms of argument and using the resources of deductive
inference to evaluate their validity. Moreover, it could be claimed (by analogy
with the perceptual case) that judgements about validity can be noninferentially justified on the basis of the deliverances of such a faculty, without
requiring any additional inferential backing.
However, there are compelling objections against assimilating our
knowledge of logical validity too closely to the perceptual paradigm. The first
is that perception of an object or an instantiated property requires the holding
of a causal relation between the perceiving subject and the object or
instantiated property. Valid forms of argument, however, are abstract objects,
and as such, are incapable of entering into causal relations. So, there is no
coherent notion of perceiving the validity of a form of argument. The second
objection is that perception alone can only yield knowledge of contingent
truths.7 But knowledge of logical validity is knowledge of necessity. Therefore
perception alone is incapable of explaining this modal dimension of our logical
knowledge. The upshot is that the notion of genuinely perceptual access to
the validity of forms of argument is incoherent; but without further elaboration,
the notion of quasi-perceptual access is merely obscure.
In addition, Paul Boghossian (2001, 2003) has argued that the appeal to
rational intuition cannot do the work required of it. Suppose we grant that
deductive inference is not required for forming a rational judgement to the
effect that a certain form of argument is valid; still, he argues, it is required in
order to bring any such judgement to bear on the rationality of a particular
inference. So, if we judge on the basis of rational intuition that any inference
of the modus ponens (MPP) form is valid, then we still need to make the
following inference:
(i)
Any inference of the form MPP is valid.
(ii)
This particular inference, from (1) and (2) to (3) is of MPP form.
Therefore,
(iii)
This particular inference from (1) and (2) to (3) is valid. (2003,
p.233)
A similar objection applies if we grant that what rational intuition provides is
not direct access to the validity of a general form of argument, but rather to
the validity of a particular instance of that form.8 The thought is that inference
6
See Boghossian (2001), (2003) for a statement of these objections.
7
See Peacocke (2003) for an account of a posteriori knowledge of necessity on which it
decomposes into a priori knowledge of necessity combined with a posteriori knowledge
of contingency.
8
The suggestion was made, in response to Boghossian (2001), by Wright (2001).
69
is still required in order to bring the judgement to bear on the rationality of
making a deductive inference in the particular case, along the following lines:
(i)
This particular inference from (1) and (2) to (3) is valid.
(ii)
If an inference is valid, then anyone who is justified in believing its
premises and knows of its validity is justified in inferring its
conclusion.
Therefore,
(iii)
Anyone who is justified in believing the premises of the argument is
justified in inferring its conclusion.
(iv) I am justified in believing the premises (1) and (2).
Therefore,
(v)
I am justified in inferring (3). (2003, p.234)
6.
Dummett on the Perception of Semantic Patterns
Despite the force of Boghossian’s objections, it seems to me that rational
intuition is both a genuine phenomenon and one to which we can usefully
appeal in giving substance to the Brewer/Peacocke account of the rationality
of deductive inference. In arguing the point, I will draw on some illuminating
remarks made by Michael Dummett on the matter.
In the following passage, Dummett draws a distinction between
understanding a proof and merely checking that it is correct:
…an understanding of a proof demands more than an ability to
recognize that it is correct. To verify that every line of a formal proof
follows from earlier lines by one of a list of transformation rules is to be
convinced, within the limits of human error, that it is correct; but it takes
one very little way towards understanding the proof. The proof has an
architecture that must be comprehended as a whole; but the first
necessity for gaining such comprehension is to be intuitively convinced,
for each step, that it genuinely follows from the earlier lines from which it
was derived. On Frege’s account, this will in general require a creative
act. It is not enough merely to grasp the thought expressed by each line
of the proof; in addition, one must perceive patterns common to those
thoughts and others, patterns which are not given with the thoughts as a
condition for grasping them but which require a further insight to
apprehend. (1991, p.197-8)
To understand the proof, it is not enough to understand each line of the proof
and to have checked that each line follows from the earlier lines by application
of the inference rules. In addition, it is required that one should perceive
certain patterns which hold between the propositions or thoughts expressed
by the various lines of the proof. This “perception of pattern” is not something
which is built into the requirements for understanding the various lines of the
proof, but requires an additional, creative insight. Thus, on Dummett’s
account, following a proof is as much a creative matter as discovering a proof.
70
Dummett contrasts the case of understanding a proof with the imaginary
case of an obedient community under the governance of an International
Academy of Logic, a body which lays down decrees concerning which logical
laws are, until further notice, to be treated as valid or invalid. He observes:
…we could obey these decrees; but we should lose the sense that we
any longer understood what we were saying. The rules of the languagegame would be clear enough; but its point would now escape us. (1991,
p.207)
Presumably, what would be lacking in such a community is not merely the
capacity for perception of pattern – its members could be perfectly capable of
perceiving the formal patterns holding between the lines of a proof – but rather
the capacity for perception of semantic patterns: for example, that the truth of
the conclusion of an argument follows from the truth of its premises. The idea
seems to be that in the absence of a capacity for representing such semantic
relations, the subject cannot have any sense of the point of his inferential
practices: he will be merely following the rules without understanding why he
is following them.
It is here that an account of the rationality of deductive inference links up
with an account of the thinker’s understanding of logical concepts, which
consists in knowledge of their contribution to truth-conditions. After all, it is
presumably in virtue of the thinker’s knowledge of the contribution made to
truth-conditions by logical concepts that he is capable of appreciating the
semantic properties of logical forms, such as the validity of a form of
argument. Dummett’s idea seems to be that what is lacking in his imaginary
community is precisely the knowledge of reference in which the understanding
of logical concepts consists:
…the missing component of understanding is not to be stigmatized as a
‘mere’ feeling…. What he lacks is not the feeling of understanding, but
the knowledge that is an essential component of understanding. It is that
knowledge that we should lack if we were compelled to reason in
accordance with principles that appeared to us invalid or gratuitously
restricted: we could rightly confess that we no longer knew what we were
saying. (1991, p.208)
7.
Meeting Boghossian’s Challenge
In summary, Dummett’s idea is that understanding a proof requires perceiving
the semantic patterns holding between the different lines of the proof or, in
other words, perceiving the validity of the proof. But how are we to understand
talk of “perceiving” in this context? We have already seen compelling
objections against literal interpretations of such talk, but perhaps it can be
understood metaphorically.9 It seems to me that the perceptual metaphors
suggest themselves because there are clear similarities between discerning
the patterns in a proof and a certain kind of perceptual phenomenon. Indeed,
9
After all, we use perceptual vocabulary in a wide variety of settings that do not involve
sensory perception, for example: seeing that a certain course of action is the one to
take, seeing that one has made a mistake in a proof, and so on.
71
it is plausibly one and the same kind of phenomenon which occurs both in the
context of perception and in the context of pure thought. This is the
phenomenon that Wittgenstein labeled “seeing as” in the second part of the
Philosophical Investigations. Just as one can be struck by certain spatial
relations between objects in the perceived array, so one can be struck by
certain semantic relations between thoughts in occurrent consciousness. In
some cases, this may involve a kind of conceptual structuring of the contents
of conscious experience, but the basic phenomenon is pre-conceptual – it is a
matter of directing one’s attention or having one’s attention directed in such a
way that the relations in question become salient in consciousness.
This basic attentional phenomenon has a number of features which
make the perceptual metaphors particularly apposite. First, it involves a
characteristic kind of passivity, which is given expression in the idea of being
struck by the relations in question. Second, and relatedly, being struck by
such relations is not something which stands in need of reasons or
justification, any more than does the passive reception of perceptual
experience itself. Third, the phenomenon is belief-independent. Just as we are
subject to cognitively impenetrable perceptual illusions, such as the MullerLyer illusion, which may persist in the phenomenology of experience even
when we know about the illusion, so we are sometimes subject to illusions of
validity which persist in the same way even when we know about them.10
These points bring out a sense in which the analogy between rational
intuition and perception need not be obscure, but can actually be quite
illuminating. They also defuse Boghossian’s threat of the reappearance of
Carrollian circularity. To the extent that the representations of logical validity
that we are considering are belief-independent, it cannot be assumed that
inference must be involved in bringing them to bear on the rationality or
justifiability of a particular inference. On the contrary, it can be argued that the
subject’s conscious attention to the fact that the conclusion of an argument
follows from its premises may be directly causally and rationally implicated in
the fact that he makes the inferential transition from a belief in the premises to
a belief in the conclusion.
So, despite Boghossian’s challenge, the notion of rational intuition may
have an important role to play in an account of the rationality of making a
particular deductive inference. In my view, it certainly plays an indispensable
role in giving a plausible account of propositional knowledge of logical truths,
including knowledge of which forms of argument are logically valid. Consider
what is involved in the rationality of judging that a particular argument of a
certain form is valid. On one view, which we can call the deductive model, this
requires giving an explicit articulation of the general form of the argument in
question and inferring its validity from more general principles concerning
what it is for a form of argument to be valid. But there is an alternative view,
which we can call the rational intuition model, on which judgements of validity
in the particular case have epistemological priority and can be rationally based
directly on the deliverances of rational intuition. On this view, one’s conscious
awareness of the fact that the conclusion of an argument follows from its
premises, which is explained by one’s understanding of the logical concepts
10
72
Ayers (1991) gives the example of Zeno’s paradoxes in discussing illusions of validity.
that figure in the argument, provides one with a non-inferential rational basis
for judging that the argument is logically valid.
The comparison with the epistemology of moral and epistemic
judgements is instructive. If we consider what is involved in the rationality of
judging that a particular case exemplifies knowledge, or goodness, then our
options are structurally similar. On the deductive model, it requires giving an
explicit articulation of the conditions that are necessary and sufficient for the
exemplification of knowledge, or goodness, and showing that these conditions
are satisfied in the particular case in question.11 On the rational intuition
model, by contrast, judgements about the exemplification of knowledge, or
goodness, in particular cases have epistemological priority and can be
rationally based directly on the deliverances of rational intuition. On this view,
one’s conscious awareness of the fact that a particular case exemplifies
knowledge, which is explained by one’s understanding of the concept of
knowledge, provides one with a non-inferential rational basis for judging that
the case exemplifies knowledge.
The cases of moral and epistemic judgements bring out particularly
clearly the intuitive appeal of the rational intuition model, as compared with the
deductive model. But its intuitive appeal is supported by considerations of
more theoretical nature. The problem with the explicit inference view is not
just that it makes the acquisition of moral, epistemic and logical knowledge an
extremely difficult and arcane matter; it also makes it viciously circular. In
order to know that a particular case exemplifies knowledge, I already need to
know a great deal about knowledge – indeed, I need to have explicit
knowledge of which conditions are necessary and sufficient for knowledge.
But how am I to acquire this knowledge? At this point, it is hard to deny that
we acquire knowledge of necessary and sufficient conditions for knowledge
on the basis of an attempt to systematize and explain the judgements that we
are pretheoretically inclined to make in particular cases, and which we are
(defeasibly) entitled to presume are in good epistemic standing. Certainly, this
has been the methodology which characterizes the mainstream in the
literature on the conceptual analysis of knowledge. But the explicit inference
view makes it viciously circular.
For similar reasons, it is quite implausible to require that I need to have
articulated, or even to be capable of articulating, the general form of an
argument in order to be rational in judging that a particular argument of that
form is valid. Consider the following “Days of the Week” argument:
(1)
Today is Monday.
(2)
If today is Monday, then tomorrow is Tuesday.
Therefore,
(3)
Tomorrow is Tuesday.
I can just see that this argument is valid, and be rational in judging that it is,
without being able to articulate the general form of the argument. It is not just
11
See McDowell (1979) for an early critique of this view. I hope in the future to discuss
Peacocke’s (2003) view that we have tacit knowledge of moral principles which are
brought to bear in making judgements in particular cases.
73
that the capacity to articulate the general form of the argument is somewhat
arcane and technical; it also depends on a more fundamental capacity to
evaluate particular arguments as valid or invalid. Suppose I form the belief
that the Days of the Week argument is valid and go on to try to justify my
belief on the basis of the claim that the argument instantiates the following
form which I also claim to be logically valid:
(1)
P
(2)
If P, then Q
Therefore,
(3)
Q
Then I am open to apparent counterexamples, in the style of Van McGee,
involving instances of this form which have conditionals embedded in the
antecedent of the conditional premise. Now, I might try to explain away the
intuitive plausibility of these apparent counterexamples and, who knows, I
might even be successful, but their initial plausibility cannot be denied. The
crucial point is that my judgement that the Days of the Week argument is valid
(as opposed to my attempt to give an explicit justification of that judgement)
should not depend on the outcome of this debate. Rather, it is plain that the
debate itself is a debate about how best to systematize and explain the
judgements we are pretheoretically inclined to make in particular cases, which
we are defeasibly entitled to presume are in good epistemic standing.
A purely deductive model of our acquisition of logical knowledge is
unacceptable; it needs to be recognized that the acquisition of logical
knowledge depends both on rational intuition and on a broadly inductive
process of inference to the best explanation of judgements based on rational
intuition.
8.
On Brewer’s Non-Reflective Access Requirement
For these reasons, then, it seems to me that rational intuition is both a
genuine phenomenon and one that has an important role to play in the
epistemology of logic. However, it is a further question whether it can be
invoked in a plausible defence of Brewer’s claim that the rationality of
deductive inference depends on the fact that the inference is made on the
basis of a conscious appreciation of its validity. Brewer’s proposal seems to
be that a subject’s belief in the premises of a deductively valid argument
provides him with a reason for believing its conclusion in virtue of embodying
an appreciation of the logical validity of the form of argument in question.
According to this line, the appreciation of logical validity is actually embodied
in the very beliefs which serve to provide the premises of the inference. But
this is not Dummett’s conception of the matter. According to him:
It is not enough merely to grasp the thought expressed by each line of
the proof; in addition, one must perceive patterns common to those thoughts
and others, patterns which are not given with the thoughts as a condition for
grasping them but which require a further insight to apprehend. (1991, p.198)
Indeed, he takes this point to be crucial in the explanation of how it is
that deductive inference can be informative, in the sense of providing a source
of new knowledge:
74
Deductive reasoning is thus in no way a mechanical process, though it
may be set out so as to be checkable mechanically: it has a creative
component, involving the apprehension of patterns within the thoughts
expressed, and relating them to one another, that are not required for or given
with a grasp of those thoughts themselves. Since it has this creative
component, a knowledge of the premises of an inferential step does not entail
a knowledge of the conclusion, even when we attend to them simultaneously;
and so deductive knowledge can yield new knowledge. Since the relevant
patterns need to be discerned, such reasoning is fruitful; but since they are
there to be discerned, its validity is not called into question. (1981, p.42)
If believing the premises of a valid argument were sufficient to provide
you with an appreciation of how the truth of the conclusion follows from the
premises, then deductive inference would not be informative. On Dummett’s
account, by contrast, it requires a creative insight to see how the conclusion of
an argument follows from its premises, and so the informativeness of
inference is preserved.
So, on Dummett’s account, it is possible to understand the premises of
an argument without thereby having any conscious appreciation of the fact
that the conclusion of the argument follows. We therefore have to consider the
case in which a subject believes the premises of an argument and makes the
transition in thought to believing its conclusion, but without being guided by
any conscious appreciation of the fact that the conclusion follows from the
premises. According to Brewer, such a transition cannot be rational. But this
seems to me to be setting the hurdle too high. Brewer and Dummett make a
convincing case that following a proof with a genuine sense of understanding
requires an appreciation of its validity. But following a proof is a relatively
sophisticated activity. There are much more mundane cases of deductive
reasoning in which it is not so plausible to demand an appreciation of validity,
or even the capacity for appreciation of validity, as a requirement for
rationality.12
9.
A More Permissive Proposal
I want to make a more permissive proposal according to which a deductive
inference is rational, in the most basic cases, just in case the form of the
inference is a valid form, and the thinker makes the inference because it is of
that form.
What is it to make an inference because it is of a certain form? I have in
mind the following kind of account. At the personal level, it is just a matter of
exercising a disposition to make inferences of that form, but at the
subpersonal level, the disposition must be grounded by some state that is a
common cause of the various exercises of the inferential disposition, and
hence which embodies tacit knowledge of the form of inference in question.13
On this account, there is no requirement on representation of the form of
inference at the personal level: in particular, there is no requirement that the
12
I have in mind cases of deductive inference in animals and children, but also in
unreflective adults and reflective adults in their more unreflective moments.
13
See Peacocke (1992) Ch.7.1 for a more detailed proposal along these lines.
75
subject should be guided in making the inference by any conscious
appreciation of the fact that the form of inference in question is a valid form.
But if this all there is to be said, then what, if anything, is left out on
Fodor’s account of the rationality of deductive inference? The suggestion we
have been considering is that deductive inference requires understanding,
which requires consciousness; but Fodor leaves out any mention of
consciousness in his account. The basic idea behind this suggestion seems to
me to be correct, but it can be deployed in different ways. The
Dummett/Brewer/Peacocke line is that making a deductive inference in a way
which involves genuine understanding requires a conscious appreciation of
the validity of the form of inference. My objection was that, while this
requirement may be appropriate in relatively sophisticated cases, such as
following a proof, it is inappropriate in more mundane cases of deductive
inference. But I think there is a different way to make the connection between
inference, understanding and consciousness.
A deductive inference is certain kind of (non-deviant) causal transition
between beliefs. And in order to have a belief with a certain content, it is
necessary to understand that content, which requires understanding its
component concepts. Now, the crucial question is this: what is required for
understanding a logical concept? And the crucial claim is that it is not
sufficient to have tacit knowledge of what are in fact valid logical forms. Tacit
knowledge of logical forms is just a matter of having a full-blooded disposition
to make transitions between intentional states whose contents instantiate
those forms. There is no requirement for understanding the contents of the
states involved in the relevant transitions. (Compare: the intentional contents
involved in low-level computations in the visual or grammar modules.)
Understanding a logical concept, on the other hand, is a matter of having a
full-blooded disposition to make deductive inferences between beliefs whose
contents instantiate certain logical forms. Since we are now in the domain of
belief and inference, possession of the relevant disposition does require
understanding the contents of the beliefs involved in the inference.
But now we seem to be moving in a circle which is too small to be
illuminating. How can we get a grip on what is distinctive about inferences
between beliefs as compared with merely inference-like transitions between
merely belief-like intentional states, without simply appealing to a notion of
understanding which is taken as primitive? How can we get some
independent grip on the claim that understanding is involved in the one kind of
case, but not the other?
The answer I propose is that the notions of belief, inference and
understanding are normatively individuated: that is to say, they are
individuated by their rational role. We can make sense of the distinction
between causally isomorphic systems of inference-like transitions between
belief-like intentional states on the one hand, and inferences between beliefs
on the other, in virtue of the fact that the latter exhibit rational relations,
whereas the former do not. But we have still not yet reached a satisfactory
stopping point. There must be some property of beliefs in virtue of which they
are capable of standing in rational relations to one another, a property which
76
might be lacked by a causally isomorphic system of belief-like states. That
property, I claim, is availability to consciousness.14
In summary, understanding a logical concept is a matter of being
disposed to make distinctively rational inferences between beliefs, where the
rationality of these inferences is grounded, in part, by the fact that the beliefs
involved are available to consciousness. So, as promised, there is an
alternative to the Dummett/Brewer/Peacocke account of the connection
between consciousness, understanding and the rationality of deductive
inference.
This alternative is more permissive than the Dummett/Brewer/Peacocke
account insofar as it denies that a rational deductive inference must be made
on the basis of a conscious appreciation of the validity of the form of inference
in question. I argued that this requirement might be plausible for relatively
sophisticated cases, such as following a proof, but not in more mundane
cases of deductive reasoning. But I also think that the permissive alternative
provides the materials for explaining how conscious appreciation of validity is
possible in more sophisticated cases. According to my proposal, the following
are requirements for the rationality of a deductive inference:
(1)
The thinker must understand the contents of the beliefs which
serve as the premises and conclusion of the inference; in other
words, the thinker must know what it would be for them to be true;
(2)
The beliefs involved in the inference must be available to
consciousness.
The fact that there are conditions under which the beliefs involved in the
inference are occurrent in consciousness grounds the possibility that the
thinker may become consciously aware of certain relations between their
contents. And given his knowledge of what it would be for them to be true, it
grounds the possibility that he may become aware of certain semantic
relations between their contents: in particular, that the truth of the conclusion
follows from the truth of the premises. In other words, it grounds the possibility
of conscious awareness of the validity of the logical form of the inference.
10. Access Internalism
Brewer’s non-reflective access requirement is motivated by the need to
accommodate the internalist intuitions that are prompted by standard
counterexamples to pure reliabilism while avoiding the problems faced by
traditional forms of access internalism. According to Brewer’s diagnosis, these
problems stem from the following pair of commitments:
(1) to the thesis that a person’s reasons are essentially recognizable by
her as such; (2) to the idea that this recognition can only be a matter of
her second-order knowledge that the mental state providing the reason
in question is appropriately related to that for which it is a reason, where
this is independent of the first-order state itself in that she might have
14
These claims are elaborated and defended in more detail in my papers, “Rationality
and the Subject’s Point of View” and “The Autonomy of Personal Level Explanation”.
77
been in just that state yet not had the second-order knowledge required
for its status as a reason for her. (1999, p.164)
The pure reliabilist response is to reject (1) altogether, whereas Brewer’s
suggestion is that we can endorse (1) while rejecting (2). Effectively, then, his
proposal is that we should replace the traditional reflective access
requirement by means of the unreflective access requirement.
However, there is another way to characterize the elusive middle ground
that we should be seeking. According to this suggestion, what is wrong with
traditional forms of internalism is not the reflective access requirement per se,
but rather the traditional interpretations of that requirement. On the traditional
interpretations, the possibility of access (i.e. recognition of reasons) is
grounded in such a way as to impose substantive constraints on the reflective
capacities of the rational subject in question. This interpretation of the
requirement is basically taken over by Brewer, except that the constraints are
imposed on unreflective, as opposed to reflective, capacities of the rational
subject. On the alternative interpretation, by contrast, the possibility of access
is grounded in the nature of the reason-giving states themselves. Thus, even
if the subject is not capable of recognizing the rationality of her belief, it may
be rational all the same, so long as it is formed in such a way as to ground the
possibility (in principle) of recognition.15
In giving an account of the requirements for rationality, we always have
to be sensitive to the fact that there are many different degrees of
sophistication in rational thought and action. We should not be so impressed
by the more sophisticated kinds of rationality that we end up denying that
more mundane kinds qualify as rational at all. But neither should we be so
impressed by the more mundane kinds that we rob ourselves of the resources
for explaining the more sophisticated kinds. I have tried to give an account
that is as permissive as possible, but which provides the resources for
explaining how it is that the most mundane kinds of rational thinking make
possible the more sophisticated kinds. Thus, it is intended to exhibit what
seems to me to be the correct methodology in this area.
Bibliography
Ayers, Michael (1991) Locke, volume 1. London: Routledge.
Block, Ned (1980) “Psychologism and Behaviourism”
Boghossian, Paul (2001) “How Are Objective Epistemic Reasons Possible?”
Philosophical Studies.
Boghossian, Paul (2003) “Blind Reasoning,” Proceedings of the Aristotelian
Society, Supplementary Volume.
Brewer, Bill (1995) “Mental Causation: Compulsion by Reason”, in
Proceedings of the Aristotelian Society, Supplementary Volume 64.
Brewer, Bill (1999) Perception and Reason, Oxford University Press.
15
78
See “Rationality and the Subject’s Point of View” for further defence of these claims.
Dummett, Michael (1981) The Interpretation of Frege’s Philosophy. London:
Duckworth.
Dummett, Michael (1991) The Logical Basis of Metaphysics. London:
Duckworth.
Fodor, Jerry (1987) Psychosemantics. MIT Press.
Fodor, Jerry and Pylyshyn, Zenon (1988) “Connectionism and Cognitive
Architecture: A Critical Analysis,” Cognition 28.
Fodor, Jerry (1998) Concepts: Where Cognitive Science Went Wrong. Oxford
University Press.
McDowell, John (1979) “Virtue and Reason,” The Monist 62.
McGee, Vann “A Counterexample to Modus Ponens”
Peacocke, Christopher (1994) “Content, Computation and Externalism”, Mind
and Language 4.
Christopher Peacocke (2003) The Realm of Reason, Oxford University Press.
Searle, John (1980) “Minds, Brains and Programs,” Behavioural and Brain
Sciences 3.
Searle, John (1990) “Consciousness, Explanatory Inversion and Cognitive
Science”, in Behavioural and Brain Sciences 13.
Searle, John (1992) Rediscovering the Mind.
Wright, Crispin (2001) “On Basic Logical Knowledge,” Philosophical Studies.
79
Consciousness and Self-Consciousness
Research Centre
www.warwick.ac.uk/go/consciousnesscentre